Major benefits of guarding behavior in subsocial bees: implications for social evolution

Abstract Parental care is a behavior that increases the growth and survival of offspring, often at a cost to the parents' own survival and/or future reproduction. In this study, we focused on nest guarding, which is one of the most important types of extended parental care; we studied this behavior in two solitary bee species of the genus Ceratina with social ancestors. We performed the experiment of removing the laying female, who usually guards the nest after completing its provisioning, to test the effects of nest guarding on the offspring survival and nest fate. By dissecting natural nests, we found that Ceratina cucurbitina females always guarded their offspring until the offspring reached adulthood. In addition, the females of this species were able to crawl across the nest partitions and inspect the offspring in the brood cells. In contrast, several Ceratina chalybea females guarded their nests until the offspring reached adulthood, but others closed the nest entrance with a plug and deserted the nest. Nests with a low number of provisioned cells were more likely to be plugged and abandoned than nests with a higher number of cells. The female removal experiment had a significantly negative effect on offspring survival in both species. These nests frequently failed due to the attacks of natural enemies (e.g., ants, chalcidoid wasps, and other competing Ceratina bees). Increased offspring survival is the most important benefit of the guarding strategy. The abandonment of a potentially unsuccessful brood might constitute a benefit of the nest plugging behavior. The facultative nest desertion strategy is a derived behavior in the studied bees and constitutes an example of an evolutionary reduction in the extent of parental care.


Introduction
Parental care is a diverse life history trait that includes various types of behaviors (Tallamy and Wood 1986;Trumbo 2012) and that influences other life history characteristics of animals (Gilbert and Manica 2010). The three primary types of parental care are provisioning with food, offspring guarding, and building nests or shelters (Wilson 1971;Thiel 2000;Smiseth et al. 2012).
Parental care is crucial for offspring survival in certain species (Smiseth et al. 2012), while it only increases the fitness of the offspring in others (Martins et al. 1998;Mas and K€ olliker 2008). Offspring are usually more dependent if the main form of care is provisioning rather than guarding (Smiseth et al. 2012). The dependency of the offspring on parental care is also affected by the age of the offspring, that is, the younger stages are more parent dependant (Coville and Griswold 1984;Eggert et al. 1998;Smiseth et al. 2003).
One of the most important features that affect the life history of species with respect to parental care is the trade-off between the length of care for the offspring and the number of offspring produced in a lifetime. Longer parental care causes a decrease in the number of offspring over time (Gross 2005;Smiseth et al. 2012;K€ olliker et al. 2015). Similarly, when offspring are produced in separate clutches, the parents must decide whether it is more effective to care for the current clutch or to desert that clutch and establish a new one (Olmstead and Wood 1990;Mas and K€ olliker 2008).
The aculeate Hymenoptera are one of the most important and most studied groups of invertebrates with parental care (Wilson 1971;Tallamy and Wood 1986;Linksvayer and Wade 2005). In nonkleptoparasitic aculeate Hymenoptera, the parental care typically consists of nest building and provisioning accompanied with nest guarding in certain species (Wilson 1971;Michener 1974). The offspring are provisioned in two primary modes in the aculeate Hymenoptera. The mass provisioners collect and supply a cell with all the necessary food for the young; then, they deposit an egg and close the cell. Typically, the adults do not interact with the offspring; therefore, guarding is not likely to be important for offspring survival (Strohm and Linsenmair 2000;Field 2005). In contrast, the progressive provisioners feed (and guard) their offspring regularly throughout larval development (Field and Brace 2004;Field 2005). Interestingly, there are a few exceptions to this scheme. For example, mothers stay with their offspring until they reach adulthood in certain dung beetles (Trumbo 2012), in eusocial mass-provisioning halictid bees (Michener 1974) and in solitary populations of Halictus rubicundus (Eickwort et al. 1996). A remarkable strategy of parental care is also found in the mass-provisioning genus Ceratina (Sakagami and Maeta 1977;.
Bees of the cosmopolitan genus Ceratina nest in dead stems or sticks with pith (Hogendoorn and Velthuis 1999;Michener 2007) and are either solitary or facultatively eusocial (Sakagami and Maeta 1977;Rehan et al. 2009). The ancestor of this genus (and of the entire Xylocopinae subfamily) was facultatively eusocial (Rehan et al. 2012). Ceratina bees guard their nests after the end of provisioning. The nest is guarded by either a single female (mother) or more than one adult female (mother and daughter) until the offspring reach maturity (Sakagami and Laroca 1971;. This guarding behavior is found in all the studied species of Ceratina bees ). In addition, the guarding behavior is not only a passive strategy in Ceratina bees. Females are subsocial and typically crawl through the cell partitions in the nest to examine the offspring (Maeta et al. 1997;. To test the importance of nest guarding, we performed an experiment where the guarding individual was removed from a nest, and the fate of the offspring in the unguarded nest was monitored. Removal experiments performed in certain invertebrates showed that the absence of the parents could result in an increased mortality of the offspring and a poorer growth (Eggert et al. 1998;Thiel 2000;K€ olliker and Vancassel 2007;Werneck et al. 2012).
For the aculeate Hymenoptera, only a few experiments in which the guarding female(s) were removed were performed, primarily with eusocial (Smith et al. 2003) or communal species (Kukuk et al. 1998). Removal experiments were further used to test the role of males in species with biparental care (Coville and Griswold 1984) or in eusocial species with extended male care (Sen and Gadagkar 2006;Lucas and Field 2011). However, no removal experiments have been performed to test the role of the female mother in solitary aculeate Hymenoptera. Moreover, there are observational studies that compared the mortality of brood cells in guarded and naturally orphaned nests, but these studies have an insufficient number of observations and provide ambiguous results (Sakagami and Maeta 1977;Eickwort et al. 1996;Rehan et al. 2009).
The lack of relevant studies that examined the fate of orphaned nests is surprising because nest failure after orphaning is often argued to be one of the most important driving forces in the evolution of sociality. The insurance of offspring survival in the case of the death of the founder represents an important selection pressure for the maintenance of eusociality (Gadagkar 1990;Queller 1994) and for the existence of pleometrotic nest founding (Queller 1994;Queller et al. 2000). The probability of offspring survival after the death of the mother is also considered to affect the benefits obtained from the progressive provisioning (Field 2005).
In this study, we tested the importance of nest guarding for mass-provisioning Ceratina bees, which are solitary species with social ancestors. We demonstrated that there are two different strategies to protect the nests, which include an alternative to nest guarding that had not been previously recognized in Ceratina bees. We studied these different nesting strategies in two relatively distantly related species: Ceratina cucurbitina (Rossi) and Ceratina chalybea Chevrier ( Fig. 1; Terzo 1998).

Study site
The observations and experiments were conducted in the Podyj ı National Park, near Znojmo, in the Czech Republic. We received permission to perform the experiments in the National Park (NPP 0781/2011). Most of the data were collected in the locality of the Havranick e v resovi st e heathland (48°48 0 33.595″N, 15°59 0 35.149″E), but additional data were collected at the Sobes locality (48°49 0 0.124″N, 15°58 0 37.708″E). The localities were both open grassy habitats, with solitary trees and shrubs. The study was performed between 2012 and 2014, from the end of May to the beginning of August. The studied species (C. cucurbitina and C. chalybea) are abundant in the studied localities. They naturally nest in broken twigs and stems of Rosa canina, Centaurea spp., Linaria genistifolia, Verbascum spp., and other plants with pith.

Preparation of nesting opportunities
All the experiments were performed on nests that were artificially prepared as nesting opportunities. Sheaves consisting of 20 cut plant stems were distributed in the study site. Each stem was approximately 30-50 cm long, with an inner pith of more than 3 mm wide in the upper end (Fig. S1). We used stems of Solidago spp., Helianthus tuberosus, Echinops sphaerocephalus, and Dipsacus sylvestris. The sheaves were fixed to a rod for stabilization, and they were then fixed to the ground and placed in suitable nesting sites. These nesting opportunities were distributed in the studied localities in April before the beginning of the nesting season. We installed 2600 sheaves, which corresponded to 52,000 nesting opportunities.

Nest dissection
All the nests selected for dissection were collected after 7 p.m. CEST. By collecting the nests early in the evening, we ensured that all the inhabitants had returned to the nest and were inside because Ceratina bees are not active at this time (Herrera 1990). The collected nests were carefully opened with a knife in the field laboratory. The original nest structure and the following parameters were recorded in all the analyzed nests: stem substrate species, presence of guarding adult female, the presence and number of living offspring, and the presence of natural enemies. We considered nests as attacked by a focal enemy when we observed such enemy species in the nest at the time of dissection. In the case of nest usurpation, we marked those nests as usurped when discarded pollen or a low number of fresh cells with eggs were observed in the nest.

Description of the guarding strategy
This study was performed in 2012 and 2013. We selected approximately 30 sheaves and dissected all the nests in them every 7-9 days from mid-June to mid-August. In total, we selected 360 sheaves for this experiment; however, not all sheaves contained nests. All the nests from the selected sheaves were dissected. Additional natural nests were collected in 2012; the substrates were Rubus spp., Artemisia spp., and Helianthus tuberosus.
Only nests with full brood and an undisturbed inner structure were used for description of guarding strategy. Nests that contained a larva or a pupa in the outermost cell in the line were considered as full brood nests (Rehan and Richards 2010a,b;Vickruck et al. 2011).
We recorded the presence and position of the guarding female.
We Comparison of guarding strategies of C. chalybea In C. chalybea, we discovered two alternative guarding strategiesnest guarding versus nest abandonment. For comparison of these strategies, we used only data from 2013 (187 nests from 114 sheaves). To compare these strategies, we recorded presence of guarding female, architecture of last brood cell (open vs. closed), length of nest, number of brood cells, number of brood cells with living offspring, and number of brood cells with parasitized offspring.

Phenology of C. chalybea
The analysis of the phenology was performed in 2013. For this analysis, 108 sheaves were selected, but only 74 contained C. chalybea nests. Each sheaf was regularly controlled between 10 May 2013 and 25 July 2013. The sheaves were controlled every 1-3 days. New nests and the content of older nests were recorded using a pocket flashlight during each control. Confirmed by a later nest dissection, we could reliably observe and determine the content of the first 4 or 5 cm of depth in the nests. The founding date of a C. chalybea nest was stated as either a) the date of the first observation of a new nest, where the presence of C. chalybea was subsequently recorded, or b) the date of the first observation of a C. chalybea female in a nest where another species of insect was the owner of the nest before a C. chalybea female was regularly observed. The date of nest plugging was the first date when a plug was observed and the female disappeared. In total, we used 133 observed nests in this analysis, of which 44 were plugged nests.

Removal of the guarding female
This experiment was performed in the nesting seasons of 2013 and 2014 (from the end of June to the beginning of August). The nests and sheaves used for this experiment were different from the nests used for the analysis of the guarding strategy. For this experiment, we selected 500 sheaves in 2013 and 700 sheaves in 2014 for the analysis, but only in some sheaves, we found nests in appropriate stage. Nests of C. cucurbitina or C. chalybea with a guarding female at the nest entrance were selected. The guarding female was pulled from the nest. The nest was visually inspected by shining a light inside the nest entrance, and only completely provisioned nests were used for the experiment. A completely provisioned nest of C. chalybea was characterized by either a visible partition or the presence of pollen or a larva at the nest entrance. A completely provisioned nest of C. cucurbitina was distinguished by a visible partition near the nest entrance.
When a nest was appropriate for this experiment, the guarding female was either removed (two-thirds of the cases) or was returned to the nest to serve as a control (one-third of the cases). To evaluate the influence of nest guarding on the nest success, we performed gradual dissections of all the experimental nests. The dissections were performed at 3, 8, or 20 days after the treatment, using one-third of the nests in each interval. Based on our preliminary results, the interval between the end of provisioning and the emergence of the first adult juveniles took approximately 20 days, which we considered as the approximate duration for the complete provisioning of nests.
The female was removed in 140 nests of C. chalybea For comparison of mortality in outermost and second outermost cells, we used only nests in which these cells were preserved (nest partitions were not disturbed). Therefore, we used only subset of nests for this analysis: 155 nest from 134 sheaves for C. chalybea and 150 nest form 124 sheaves in C. cucurbitina.

Comparison between unguarded C. chalybea nests and nests with the female removed
We performed an analysis for evaluating the proportion of dead offspring in unguarded C. chalybea nests to decide whether it was the cause or the consequence of nest abandonment by the female. For this analysis, we used nests of C. chalybea with undisturbed structure from 2013 (50 nests from 45 sheaves) and unguarded full brood nests, which we observed and for which we noted a plug date. The nests were dissected within 30 days from plugging (22 nests from 16 sheaves).

Data analyses
The statistical analyses were conducted in the program R 3.1.0 (R Development Core Team, 2011). When we have more nests from one sheaf, we used mean values per sheaf in all linear models or generalized linear models.
We performed these analyses: 1 Association between guarding and nest architecture of C. chalybea: chi-square test.

Results
Guarding strategy of C. cucurbitina Full brood nests of C. cucurbitina were guarded by the mothers in 58 of the 60 examined nests (96.6%). The cell partitions in the nest of this species were relatively more fragile than those in the nests of C. chalybea. The adult females were occasionally observed inside the nests around the brood cells as if they were inspecting them (5.1% of the nests with the owner female, 6 of 117 nests). This behavior agrees with the observation that excrements typically accumulate at the bottom of the nest. Moreover, no case of social nesting and male presence in the nests was recorded; however, we observed females feeding adult offspring with pollen.
Guarding strategy of C. chalybea Ceratina chalybea had two alternative types of nest protection (Fig. 2, Table 1). The nest was either guarded by females sitting by the outermost cell or was left unguarded and closed with a plug formed with pith fillings. This nest plug was similar to the material used for the nest partitions, but it was thicker and was approximately 1.5 cm in length. A strong association between the architecture of the outermost cell and the presence of a guarding mother was detected (chi-square test, v 2 = 56.8228, df = 1, P < 0.0001, Table 1). These differences in nest architecture did not reflect the stage of ontogeny of the nests. All the stages of offspring (larva, pupa, or young adult) were found in the outermost cells in both types of architectures. Therefore, the nest plugging represented an alternative guarding strategy in this species.
A mother of C. chalybea was never observed inside the nest beyond the outermost cell partition. Excrements of larvae remain inside the brood cells, and they are not removed to the bottom of the nest. The cell partitions remained undisturbed until the offspring matured and were more compact and thicker than the partitions of the C. cucurbitina nests. No case of social nesting was recorded, but a male was found in the nest entrance of full brood nests in a few cases (3%, 6 nests of 201).

Phenology of nest founding and plugging in C. chalybea
Ceratina chalybea founded new nests predominantly between May and mid-June; however, a lower frequency of nest founding was observed in the second part of June and throughout July (Fig. S2). The nest plugging began by mid-June and had the maximum occurrence between June 20 and 23. The plugging continued through the last week of June and throughout July with a lower frequency. After mid-June, the numbers of newly founded and plugged nests were similar (Fig. S2).
There was no significant difference between guarded and plugged nest in date of nest founding (binomial GLM, deviance = 0.2226, df = 1, P = 0.6371). However, late-founded nest (after June 20) were plugged very rarely.
The probability of nest failure increased with the time spent without a guarding female (Fig. 4, Table 2). Twenty days after the experimental treatment, all the offspring died in 21.6% (8 of 37) of the control nests, while this percentage was 81.4% (57 from 70) in the nests where the female C. cucurbitina was removed. All the offspring died in 3.9% (1 of 26) of the control nests and in 53.1% (25 of 47) of the nests where the C. chalybea female was removed (Fig. 4).
The probability of offspring survival in the outermost cell (unclosed cell with guarding female) was significantly affected by the removal of the female in C. chalybea (binomial GLM, deviance = 50.62, df = 1, P < 0.0001, Table 2). In contrast, the mortality of the second outermost cell (standard cell without contact between mother and offspring) was not significantly affected by the female removal in C. chalybea nests with an undisturbed structure (binomial GLM, deviance = 1.81, df = 1 P = 0.2456; Table 2). Conversely, in C. cucurbitina, the mortality of the outermost and the second outermost cell was affected by the treatment ( Table 2).

Reasons of failure of unguarded nests
The most important natural enemies were the chalcidoid parasitoids, predatory ants, and other bees that usurped the nests (Figs 5 and 6). In general, the occurrence of these enemies was more frequent and had a stronger effect on C. cucurbitina than on C. chalybea. All natural enemies were more common in nests with guarded female than in control nests both in C. cucurbitina and in C. chalybea (Fig. 6, Table S1), but differences were not always significant. In C. cucurbitina, significant difference was in the presence of chalcidoid wasps (binomial GLM, deviance = 10.53, df = 1, P = 0.0012; Table 2) and ants (binomial GLM, deviance = 11.30, df = 1, P = 0.0009; Table 2). Difference in usurpation frequency was not significant (binomial GLM, deviance = 1.2, df = 1, P = 0.2741; Table 2). On the other hand, in C. chalybea was significant difference only in nest usurpations (binomial GLM, deviance = 6.18, df = 1, P = 0.0129; Table 2), but not in the presence of chalcidoid wasps (binomial GLM, deviance = 0.57, df = 1, P = 0.4516; Table 2) and ants (binomial GLM, deviance = 2.17, df = 1, P = 0.134; Table 2). We recorded the presence of other natural enemies; yet, their frequency was too low to be statistically assessed. The numbers of all the recorded natural enemies are shown in Table S1.

Comparison between plugged nests and nest with removed female
We compared the mortality of brood cells between nests with removed females and natural plugged nests in C. chalybea, and the results showed that there was marginally significant difference (binomial GLM, deviance = 2.81, df = 1 P = 0.0932; Table S2). Difference in proportion of cells parasited by chalcidoid wasps was not significantly differ (binomial GLM, deviance = 1.31, df = 1, P = 0.2524).

Role of nest guarding for the offspring survival
The guarding of a completely provisioned nest by a parent is obviously highly beneficial for the offspring because it increased the offspring survival in both species of Ceratina bees. The primary reasons for the destruction of unguarded nests were the different natural enemies, which were significantly more common in unguarded nests than in control nests of C. cucurbitina. In C. chalybea nests, the enemies were also more common in nests with the female removed; however, only the usurpation of the nests by other bees had a statistically significant effect. Natural enemies, which cause predation and parasitism pressure, are important drivers in the evolution of extended parental care in general (Wilson 1975). We attribute the observed differences between species to a different nest structure and a different guarding strategy in each species.
The positive effects of nest guarding on offspring survival are known in other nest-making Hymenoptera; however, the current evidence is scarce. In a study by Coville and Griswold (1984), nests with a complete brood that were unguarded by a male of Trypoxylon superbum were destroyed by ants; however, the number of unguarded nests was extremely low (only two nests). The survival of the offspring was also significantly reduced after the removal of guarding female(s) in the eusocial Megalopta genalis (Smith et al. 2003) and in the communal Lasioglossum hemichalceum (Kukuk et al. 1998). The offspring survival was significantly correlated with the presence of a guarding female in the solitary populations of Halictus rubicundus (Eickwort et al. 1996). However, this latter study was only observational, and the absence of a female can be the consequence, not the cause, of the offspring death. This outcome is similar to the results of the low offspring survival and other tested features in the plugged C. chalybea nests of our study. The three published studies on Ceratina bees show ambiguous results. While the study of Sakagami and Maeta (1977) on Japanese Ceratina species showed a higher brood cell mortality in orphaned nests, the study of Rehan et al.  (2009) on C. accusator found no effects of the female disappearance. However, only a small sample size was used in the latter study. Daly et al. (1967) observed a higher attack of chalcidoid wasps but not Ichneumonidae in orphaned nests than in guarded nests. In addition, all these studies were based on only observations from nest dissections, and they did not register the number of total destroyed nests because they were not considered as Ceratina nests. According to the literature, the primary reason for the failure of unguarded nests is the occurrence of ant raids (Coville and Griswold 1984;Kukuk et al. 1998;Smith et al. 2003). Likewise, ants are most likely the primary reason for the nest failure of Ceratina bees. Furthermore, the effect of ant assaults was probably underestimated in our data because the ants moved away after their raid and left the nest completely cleaned from debris and cell partitions, which makes their detection difficult. Although we used nests in artificial nesting opportunities for our experiments, we assume that there are no differences with local natural nests in the general pattern and effect of enemies. Our artificial nests were in the immediate vicinity of natural nests, and bees and their enemies were not introduced. The observed enemies were those that usually attack Ceratina nests (Daly et al. 1967;Sakagami and Maeta 1977).
Comparison of the guarding strategy in C. cucurbitina and C. chalybea The strategy of nest defense is different between C. cucurbitina and C. chalybea, where guarding was obviously more important for C. cucurbitina. Interestingly, C. cucurbitina exhibited an obligate nest guarding and subsocial behavior, which consisted of crawling through the nest, as described in detailed studies of other Ceratina species ). This behavior most likely protects the nest from various natural threats such as parasitism by chalcidoid wasps.
In contrast, C. chalybea females guard the nest only facultatively and choose between two alternative strategies, that is, they either guard the nest or fill the nest entrance with a plug and desert it. The guarding strategy is closely associated with nest architecture; guarded nests have an unclosed outermost cell in almost all the cases and, therefore, the mother is in physical contact with last offspring. In contrast, unguarded nests have a filling plug in almost all the cases (Table 1, Fig. 2). We exceptionally observed a few guarded nests with a filling plug; however, these nests were probably recently completed, and the females had not finished filling the plug and had not yet left the nest. We also observed a few unguarded nests with an unclosed outermost cell. Such nests were probably abandoned due to the death of the mother.
To the best of our knowledge, the described nest deserting behavior in C. chalybea represents the first example of an alternative guarding strategy within the genus Ceratina. Although unguarded nests were previously found in other species, these nests were most likely orphaned after the death of the mother (Sakagami and Maeta 1977;Rehan et al. 2009). We never observed C. chalybea females crawling through nest partitions, and the larval excrements were always left in the cells. This pattern shows that females of C. chalybea, in contrast to other species, cannot crawl through the nest partitions and, therefore, cannot be in physical contact with the offspring (except for the outermost one). The higher offspring mortality in plugged nests than in guarded nests and the insignificant difference in offspring mortality between plugged nests and nests with a female removed suggest that females may also be effective against enemies. However, removing the guarding female did not have a significant influence on the mortality of the second outermost (closed) cell and on preventing the chalcidoid parasitation.
The guarded nests of C. chalybea have a relatively unique nest structure compared to those of other species. While the guarding female of C. cucurbitina (and other Ceratina species with known nest structure) sits on the nest entrance separated from the outermost offspring by a cell partition, the outermost cell in the guarded nests of C. chalybea is open, which enables a direct contact between the mother and the offspring in the outermost cell. However, no contact between the mother and the offspring in the internal cells is possible. For this reason, we consider C. chalybea as partially subsocial because of the direct contact of a female with a single offspring in a guarded nest, even though no contact occurs with internally positioned offspring or with offspring in plugged, deserted nests. In general, the offspring in the open cell suffered less mortality than offspring in the internal, closed cells. For instance, it was never parasitized with chalcidoid wasps when the guarding female was present.
In general, the guarding strategy of C. cucurbitina is more effective for the offspring survival; however, in the case of the absence of the female, the offspring are more vulnerable than in C. chalybea.

Benefits and costs of guarding and deserting nests
The benefit of nest guarding is an apparently high offspring survival. We suppose that the benefit of deserting a nest is the possibility of founding a second nest elsewhere. We did not observe the fate of females that plugged and deserted nests. However, they most likely tried to found new nests. Females that plugged their nests finished them from mid-June to July 2013. In this period, the frequency of new nest founding was similar as frequency of newly plugged nests (Fig. S2). Females that founded nests in July usually had damaged wings (M. Mik at, K. Cern a and J. Straka, unpubl. obs.), which indicates that they had already been highly active and had probably founded a nest elsewhere. There was not any significant difference in date of nest founding between guarded and plugged nests, which shows that probably only some females deserted from their first nest and try to found the second nest.
We found that guarded nests had a significantly higher number of provisioned cells and live offspring than plugged nests (Table 1, Fig. 3). The lower number of provisioned cells in plugged nests indicates that C. chalybea females deserted less valuable nests with a higher probability. To terminate the investment in an unpromising offspring or clutch is an adaptive behavior known in other animal species (Olmstead and Wood 1990;Manica 2002). However, a question remains to be answered: Can a C. chalybea female assess the survival of her offspring and desert nests with higher offspring mortality or is the higher mortality only the result of the deserting strategy? The nonsignificant differences in offspring mortality and parasitism by chalcidoid wasps between plugged nests and nests with the removed female suggest that a higher mortality in plugged nests in comparison with guarded nests is the result rather than the cause of deserting.

Evolution of the extent of parental care in Ceratina bees
Parental care typically tends to increase in complexity (Smiseth et al. 2012;Trumbo 2012). However, Ceratina bees are an example of a reduction in the extent of parental care throughout their evolution. Eusociality was lost in certain lineages of the genus Ceratina (Rehan et al. 2012) and, at least in C. chalybea, the extended maternal care further decreased by the reduction of the nest guarding activities and the loss of the ability to crawl through the cell partitions to inspect the brood.

Data Accessibility
All the primary data were uploaded and are available as online supporting information. of Ceratina bees. We are grateful to the office of the Podyj ı National Park for the permission to conduct the research and for their helpfulness. The Grant Agency of Charles University in Prague (GAUK 574812) and the SVV project (Integrative Animal Biology) no. SVV 260 313/2016 financially supported this research.

Supporting Information
Additional Supporting Information may be found online in the supporting information tab for this article: Figure S1. Installing of artificial nesting opportunities in the Havran ıky heathland (A). Figure S2. Phenology of nest founding and nest plugging in C. chalybea in season 2013. Table S1. Recorded natural enemies of Ceratina bees in nests with removed female and in control nests. Table S2. Results of binomial GLM models comparing nests with removed females and plugged nests.