1 Introduction

Biologists were always fascinated by social insect societies (Keller and Chapuisat 2010). The origin of eusociality is one of the key puzzles of evolutionary biology (Wilson and Hölldobler 2005; Keller and Chapuisat 2010; Kennedy et al. 2017), and it was one of the problematic points for Darwin´s original theory. Although social insects are highly studied, the research has focused mainly on a few species. These species are usually highly social; therefore being useful for studying many questions concerning cooperation and conflict between individuals, however, they are unsuitable for the study of the origin of eusociality.

With their behavioral variability, facultatively eusocial species are a unique opportunity to take a look at both sides of sociality. These species are able to manifest the solitary strategy as well as the eusocial one (Schwarz et al. 2007; Shell and Rehan 2017). The largest lineages of socially polymorphic bees are Halictidae, Xylocopinae and Euglossini (Schwarz et al. 2007; Shell and Rehan 2017).

Emergence of social colonies in facultatively eusocial species can be dependent on ecological conditions, e.g. climate (Field et al. 2010; Purcell 2011). Temperate climate has often too short a season for the development of two subsequent generation of bees; therefore some species perform only solitary nesting despite retaining the ability for social nesting (Field et al. 2010). Eusocial colonies of facultatively eusocial species consist only of a few females (from two to lower tens) (Schwarz et al. 2007; Turillazzi 2013). We can compare solitary and social colonies and evaluate the success of these strategies. We might think that social colonies should have a higher productivity per female than solitary colonies, however, published evidence show that the productivity of social colonies per female is often lower than in solitary colonies (Michener 1964; Richards 2011; Rehan et al. 2014). For a proper assessment of costs and benefits of sociality, it is necessary to take the probability of nest failure into account, which is lower in social colonies (Stark 1992; Clouse 2001; Smith et al. 2007; Leadbeater et al. 2011).

Facultatively eusocial species were previously commonly considered as solitary, because sometimes it is not easy to find evidence for sociality in such species. Their frequency can be either very low in the population (Sakagami and Maeta 1987) or sociality can even occur only in some populations (Field et al. 2010) or during a limited time of a year (Dew et al. 2018; Mikát et al. 2021a).

Knowledge about the natural history of facultatively eusocial groups is essential for identifying key factors influencing the origin of eusociality. Origin of eusociality can be influenced by ecological pressures and relatedness among individuals (Keller and Chapuisat 2010; Purcell 2011; Kocher and Paxton 2014). However, the evolution of eusociality is strongly influenced by the natural history traits of each taxon (Andersson 1984; Ross et al. 2013). For example, facultatively eusocial groups with progressive provisioning such as Stenogastrinae wasps and Allodapine bees have not reversed to strictly solitary behavior (Chenoweth et al. 2007; Turillazzi 2013). On the other hand, mass provisioning groups such as Halictid bees or carpenter bees from the genera Ceratina or Xylocopa frequently reverse the strategy to strict solitarity (Rehan et al. 2012; Gibbs et al. 2012; Groom and Rehan 2018). Another example is the architecture of nests in Xylocopa. There are different tendencies in groups with linear and branched nests in Xylocopa. Species with linear nest building have the tendency to nest eusocial, but species with branched nest building have the tendency to build communal nests (Steen and Schwarz 2000). While the natural history of facultative eusocial taxa is the important knowledge for understanding the social evolution, it is still insufficient or unknown for most of the taxa. Small carpenter bees (Ceratina) are understudied in nesting and social biology compared to other facultatively eusocial species.

Small carpenter bees (Ceratina) belong to the subfamily Xylocopinae in the family Apidae and they are an exemplary taxon for the study of facultative eusociality (Sakagami and Maeta 1977; Hogendoorn and Velthuis 1999; Groom and Rehan 2018). Facultative eusociality is a primary state and the strictly solitary strategy is considered a reversion from facultative eusociality (Rehan et al. 2012). Ceratina bees nest in broken twigs. They build linear nests containing several brood cells (Michener 1990) which are mass provisioned (Rehan 2020). When provisioning is finished, the mother usually guards her offspring until adulthood (Sakagami and Maeta 1977; Rehan and Richards 2010; Mikát et al. 2016) and after that feeds mature offspring with pollen and nectar (Sakagami and Maeta 1977; Mikát et al. 2017; Rehan 2020) .

Obligatory eusocial Ceratina species do not exist. On the other hand, some reversions to strict solitarity are known (Michener 1990; Rehan and Richards 2010; Groom and Rehan 2018). Although most species are able to develop eusocial colonies, their proportion in the population is small—between 1 and 30% (Sakagami and Maeta 1984; Okazaki 1987; Rehan et al. 2010, 2015). Generally, tropical species have a higher tendency to manifest sociality than temperate species (Groom and Rehan 2018).

Social colonies are very small and usually contain only two females (Okazaki 1987; Rehan et al. 2010, 2015; Rehan 2020). Sometimes, there can be up to four residents (Sakagami and Maeta 1984; Michener 1990). More than one female in a colony can lay eggs, though usually one is reproductively dominant (Sakagami and Maeta 1984; Hogendoorn and Velthuis 1999).

There is a high diversity of Ceratina bees in the Meditarrean region with 35 species belonging to five subgenera (Terzo 1998; Terzo and Rasmont 2011). However, nesting biology and sociality in most of these species is unknown especially in comparison to North American and Japanese species (Sakagami and Maeta 1977, 1987; Vickruck et al. 2011; Lawson et al. 2018). The goal of our study was to increase our knowledge about the sociality of Meditarrean species. We performed a study in Cyprus, because eight Ceratina species live there and C. cucurbitina, which dominates in most of the Mediterranean region, is absent (Terzo and Rasmont 2011; Varnava et al. 2020).

Ceratina behavior strongly differs between different nesting phases. First, a female excavates a burrow (new founding nests), later she provisions brood cells (active brood nests) and after that she guards the offspring until adulthood (full brood nests) and lastly feeds mature offspring (full-mature and mature brood nests)(Mikát et al. 2021b). Individuals also use old nests or newly excavated burrows for overwintering (Mikát and Straka 2021). Active brood nests and full brood nests are useful for the evaluation of social status (Rehan et al. 2015). At these stages, juveniles in brood cells are present together with adults. We considered the nest social, when multiple females were present in active or full brood nest stage, consistently with previous works on Ceratina bees (Sakagami and Maeta 1977, 1984, 1987; Okazaki 1987; Rehan et al. 2009, 2015; Groom and Rehan 2018).

We discovered multifemale nests in five species and examined the frequency of multifemale nesting and the number of females in multifemale nests in each species. Moreover, we examined the difference of nest productivity and architecture between solitary and multifemale nests in three species where multifemale nests were most commonly present.

2 Methods

Study area

All samples were collected in Cyprus. Most of the samples were collected around Lefkara village (34.8744575N, 33.3113722E) and between villages Mathiatis and Agios Theodoros. However, other nests were collected in other parts of Cyprus, specifically around Paphos and Limassol. Coordinates of all sampled nests are showed in Online resource 2.

Sampling process

We sampled nests in four periods through the nesting season: 25–30 May 2018, 15–28 June 2019, 6–12 August 2019 and 17–22 September 2019. We collected nests from semi-natural nesting opportunities, such as cut or otherwise disturbed twigs next to roads. However, nests were relatively scarce in the landscape. Therefore, we cut additional twigs to increase nesting opportunities. We performed the cutting of twigs in March, June and August 2019. The most common nesting substrate was blackberry (Rubus spp.), and nests were also frequently collected from fennel (Foeniculum vulgare) and a giant fennel (Ferula communis).

Dissecting of nests

Broken twigs containing a possible nest entrance (presented as a hole in the disturbed area) of Ceratina were collected. It is not possible to determine whether the nest truly is present inside of the twig; therefore we collected all twigs which could potentially contain a nest and the presence of nest was confirmed during the dissection process. We collected nests after 6 p.m. to ensure that all inhabitants were back from foraging trips and therefore inside nests. The nest entrance was closed by a tape forthwith of collection. Nests were opened with clippers. We noted these nest features: number, sex, and age of adult individuals, number of brood cells, number and stage of development of immature offspring, presence of parasites, and nesting substrate. We measured the following parameters of nest: length of nest, length of entrance burrow, width of nest, width of pith and width of twig. In total, we sampled 20 nests of C. bispinosa Handlirsch, 338 nests of C. cypriaca Mavromoustakis, 192 nests of C. dallatoreana Friese, 131 nests of C. chrysomalla Gerstaecker, 203 nests of C. mandibularis Friese, 102 nests of C. parvula Smith and 45 nests of C. schwarzi Kocourek.

There is no existing taxonomic key for species identification of Cypriot Ceratina. We used publications from Daly (1983) and Terzo (1998) for species determination. Collected individuals were also compared to determined specimens deposited in museum in Linz, Austria.

Collected nests were in different stages: burrows, which contained only adult(s) without any sign of present or past brood cells. These burrows are either newly founded nests, burrows prepared for overwintering or temporary shelters only. However, in the case of C. dallatoreana, we distinguished burrows with nectar storage at the bottom of the nest and standard burrows without this storage. Other types of nests were active brood nests, which contained brood cells and the outermost brood cell contained egg or was recently provisioned with pollen supplies; full brood nest, which contained larva or pupa in the outermost brood cell and all brood cell partitions were preserved; full-mature brood nests, which contained immature stages and also newly enclosed adults which damaged brood cell partitions; and mature brood nests which contained only mature offspring.

Analysis of phenology

Nests which contained at least one adult female were included in this analysis. Therefore, nests with only males, juveniles or parasites were not included in the analysis because the goal of this analysis was to evaluate the behavior of females through the season. We calculated the proportion of different nest stages.

Comparison between solitary and social nests

We considered multifemale nests as social and single-female nest as solitary. We included only active brood nests and full brood nests in this analysis. We excluded orphaned nests (nests without a female), because these nests cannot be classified in any category as no female is present there. For most analyses, we analyzed a pooled dataset with those with active or full brood nests, because there was no significant difference in the number of brood cells in most species between active brood nests or full brood nests (see Results). The number of active and full brood nests together was 11 for C. bispinosa, 77 for C. chrysomalla, 164 for C. cypriaca, 65 for C. dallatoreana, 115 for C. mandibularis, 67 for C. parvula and 14 for C. schwarzi. For the analyses the number of brood cells or empty cells for were used for calculations; we excluded nests parasitized by ichneumonid or Gasteruption parasites because these parasites damaged the nest and the number of brood cells was therefore unknown. The statistical analyses were performed in R software 3.6.1. (R. Core Developmental Team 2016). Chi square tests were used for testing the proportions of social nests in active brood nests vs. full brood nests and between different parts of the season. Wilcoxon tests were used for testing the effect of sociality on number of brood cells, number of brood cells per female and length of nests, because data were not normally distributed. The effect of sociality on the proportion of empty cells was used by generalized linear model of binomial family.

Morphological characteristics of females

We measured head width as the distance between the two widest points on the head. We dissected female abdomens and extracted ovaria. Ovarian development was measured as the sum of length of the three largest oocytes (Rehan et al. 2010). Wing wear was evaluated by the relative score developed by Mueller and Wolf-Mueller (1993). This relative score recognizes seven levels of wear. When wear was between levels described by Mueller and Wolf-Mueller (1993), we evaluated it as a half-level wear. Measurements were performed on a Nikon SMZ 745 stereomicroscope.

We successfully performed measurements only for C. mandibularis, C. cypriaca and C. chrysomalla, because C. parvula and C. bispinosa were too small to successfully perform the dissection. Unfortunately, dissection was also unsuccessful in some individuals from analyzed species due to the damage to ovaria or unsuccessful location.

3 Results

3.1 Phenology

C. cypriaca

Proportion of active brood nests and full brood nests was relatively stable throughout the nesting season. The most different month was September, when the proportion of active brood nests was the lowest and the proportion of mature brood nests the highest (Figure 1).

Figure 1.
figure 1

Phenology of Ceratina species in Cyprus.

Full size image

C. ballatoreana

The proportion of active brood nests was high in May (40%, 8/20) and June (54%, 15/28), it decreased in August and no active brood nests were found in September. The main peak of full brood nests was in June and August. The proportion of burrows without nectar storage was stable throughout the season. However, the proportion of burrows with nectar storage was highest in September (77%, 54/70) and relatively common in August and May. No burrows with nectar storage were found in June.

C. chrysomalla

Proportion of active brood nests was relatively stable between May and August. However, no active brood nests were found in September. The highest proportion of full brood nests was in June and September.

C. mandibularis

Proportion of active brood nests and full brood nests was stable throughout the nesting season. Mature brood nests and full-mature brood nests occurred from June, but their proportion was stable between June and September.

C. parvula

Active brood nests were very common in May (58%, 7/12) and August (68%, 15/22), and their proportion was the lowest in September (10%, 2/21). Full brood nests occurred from June to September and their proportion was highest in September.

C. bispinosa and C. schwarzi

Not enough data.

3.2 Number of brood cells provisioned

The average number of brood cells provisioned was 4.11 (N=9, SD=3.25, range 1–8) for C. bispinosa, 1.67 (N=129, SD=1.41, range 1–9) for C. cypriaca, 3.42 (N=49, SD=2.17, range 1–11) for C. dallatoreana, 4.49 (N=63, SD=2.44, range 1–11) for C. chrysomalla, 2.79 (N=97, SD=1.96, range 1–11) for C. mandibularis, 5.11 (N=65, SD=3.66, range 1–20) for C. parvula and 2.9 (N=10, SD=1.60, range 1–5) for C. schwarzi. There was no significant difference in the number of brood cells provisioned between active brood nests and full brood nests for most species (Table I). Significant difference in brood cells provisioned between active brood nests and full brood nests was in C. cypriaca (Table I), where active brood nests had a higher number of brood cells provisioned (2.50, N=40, SD=2.11, range 1–9) than full brood nests (1.30, N=89, SD=0.66, range 1– 4).

3.3 Evidence for social nesting

We found five nest categories: solitary nests (one adult female present), multifemale nests (at least two females present), nests where adult male and female were present, orphaned nest (no adult female was present) and nests where one living adult female and one dead adult female was present. Here, we note only the proportion of multifemale nests, and number of nests of all categories are summarized in Table II for each species.

C. bispinosa

Proportion of multifemale nests was 27.27% (3/11, Table II). Other nests were solitary. Two multifemale nests contained two females, one contained three females and two males. Multifemale nests were found in May, June and September.

C. cypriaca

Proportion of multifemale nests was 6.21% (10/161). There was no significant difference in the proportion of social nests between active brood nests and full brood nests (Chi square test, X^2 = 2.20, df = 1, p = 0.13, Table III). Nine multifemale nests contained two females, one contained two females and one male . Multifemale nests were found in June, August and September.

C. dallatoreana

We found only solitary or orphaned nests (N=65).

C. chrysomalla

Proportion of multifemale nests was 3.07% (2/65). Other nests were solitary. Multifemale nests contained two females.

C. mandibularis

Proportion of multifemale nests was 18.01% (20/111). There was a significant difference in the proportion of multifemale nests between active brood nests and full brood nests (Chi square test, X^2 =17.57, df = 1, p = 2.759e-05). In active brood nests, the proportion of multifemale nests was 35.29% (18/51), and in full brood nests, it was only 3.33% (2/60). Most of the multifemale nests (70%, 14/20) contained two females. However, we also found one nest with three females, four nests with four females, and one nest with five females. Multifemale nests were found in all studied periods, proportion of social nests did not significantly differ between periods (Chi square test, X^2 = 0.51, df = 3, p = 0.91).

C. parvula

Proportion of multifemale nests was 6.15% (4/65). Other nests were solitary. Two multifemale nests contained two females, one contained four females, and one contained six females. Multifemale nests were found in August and September.

C. schwarzi

We found only solitary nests (N=14).

3.4 Differences between solitary and multifemale nests

Number of brood cells

Number of brood cells provisioned was significantly higher in multifemale nests in all three species with enough nests (Wilcoxon test: C. mandibularis: W = 1273.5, p = 4.068e-06, C. cypriaca: W = 879, p = 8.603e-05, C. parvula: W = 243, p=0.0009, Figure 2, Table IV). However, the number of brood cells per female (per capita productivity) was not significantly different between solitary and social nests in any species with sufficient sample size (Wilcoxon test, C. mandibularis: W = 735.5, p = 0.75, C. cypriaca: W = 607, p = 0.43, C. parvula: W = 134.5, p = 0.74).

Figure 2.
figure 2

Comparison of productivity of solitary and social nests. Effect is significant for all three species (Wilcoxon test: C. mandibularis: W=1273.5, p=4.068e-06, C. cypriaca: W = 879, p=8.603e-05, C. parvula: W = 243, p=0.0009).

Full size image

Length of nest

Social nests were significantly longer than solitary nests in C. mandibularis (Wilcoxon test, W = 1147.5, p = 0.035, Table IV) and C. parvula (Wilcoxon test, W = 212, p = 0.015), but not in C. cypriaca (Wilcoxon test, W = 937, p = 0.16).

Proportion of empty cells

Proportion of empty cells was higher in solitary nests (29%, Table IV) than in multifemale nests (17%) in C. mandibularis. The difference was statistically significant (GLM binomial, deviance = 13.53, df = 94, residual deviance = 94.11, p = 0.0002). However, this difference was not significant in C. cypriaca (GLM binomial, deviance = 0.59, df = 126, residual deviance = 105.54, p = 0.44) or C. parvula (GLM binomial, deviance = 1.05, df = 63, residual deviance = 18.75, p = 0.30).

3.5 Reproductive status of females

C. chrysomalla

We analyzed females from two social nests. Females in social nests differed in ovarian development in both nests (in the first nest, sum of the three largest oocytes was 4.63 mm for the first female and 1.53 mm for the second female, in the second nest the sum of three largest oocytes was 6.86 mm for the first female and 1.55 mm for the second female). Females with higher ovarian development had higher wing wear in both cases; however, in one nest, the larger female had a larger ovarian development and in other smaller female had larger ovarian development.

C. cypriaca

We analyzed females from seven nests, and all contained two females. The female with the larger ovarian development had the sum of the three largest oocytes 2.38 mm on average (N=7, SD = 1.89, range 0.95–6.15), the female with the smaller ovarian development had the sum of the three largest oocytes 1.22 mm (N = 7, SD = 0.66, range 0.56–2.36) on average. The difference was not statistically significant (Wilcoxon test, N = 14, W = 37, p = 0.12).

Females with larger ovarian development had larger wing wear in four cases, smaller wing wear in two cases, and in one case there was no difference in wing wear between females. Head width was not significantly different between females with larger and smaller ovarian development (paired t-test, N = 7 nests, t = 0.72, p value = 0.49).

C. mandibularis

All multifemale nests contain one female with extensive wing wear (score 1 and larger) in maximum. However, 27.77% (5/18) of nests contained only unworn females, and two other nests contained females from which one has only a slight wing wear (score 0.5).

Almost all females in social nests had at least some ovarian development. Ovaries usually differed between females in multifemale nests, although some nests contained multiple females with highly developed ovaries. The female with the most developed ovaries was usually the female with the largest wing wear. In 83.33% (10/12) of nests where females differed in wing wear, the female with the most developed ovaries had the highest wing wear. On the other hand, the largest female in the nest did not usually have the biggest ovarian development. In five nests, females did not differ in wing wear. In 29.41% (5/17) nests, the female with the largest head width also had the largest ovarian development, in 35.29% (6/17) of nests another female had the largest ovarian development and in 35.29% (6/17) of nests there was no difference in head width between females. Within nest variance in ovarian development was not related to the variance of within nest variance of wing wear (linear model, N = 17, F = 0.94, p = 0.35) nor head width (linear model, N = 17, F = 0.27, p = 0.61).

3.6 Branched nest architecture

In C. cypriaca, we found one nest (out of 338 nests) with two branches within the nest. This nest was solitary and contained brood cells in both branches. It was present in the stem of Ferula with 2.5 cm in diameter. C. cypriaca nests were frequently found in Ferula—the proportion of nests in Ferula was 68.64% (232/338). Although Ferula stems had 1.75 cm in diameter on average, we did not encounter more branched nests.

4 Discussion

4.1 Presence of sociality

Although Ceratina bees were traditionally considered as solitary and most nests are indeed solitary, the possibility of multifemale nesting was detected in multiple studied species (Michener 1974, 1990, 2007; Sakagami and Maeta 1977; Rehan et al. 2010). The most recent work reviewing the presence of sociality in the genus Ceratina (Groom and Rehan 2018) states that social nesting was present 26 out of 34 studied species. However, European species studied to this date are considered to be solitary or biparental (Mikát et al. 2016, 2019a, b). The only exception is C. chalybea, where unusual sociality with males as helpers occurs (Mikát et al. 2021a). Social biology of southern European species was mostly unexplored to this date. We examined the social strategy in seven species, and we detected social nests in five of them: C. mandibualis, C. cypriaca, C. chrysomalla, C. parvula and C. bispinosa. We have not detected social nests in two species: C. dallatoreana and C. schwarzi. However, the number of nests analyzed is too low to interpret the social status in C. schwarzi. For C. dallatoreana, the number of examined nests (56) can be insufficient if sociality is very rare, though Daly (1966) also reported C. dallatoreana as a solitary species. In our opinion, C. dallatoreana can be obligatory solitary, although more data from multiple regions would be useful.

Our results show that social nesting of Ceratina is also more common in Europe than previously presumed and probably most of the species can be facultatively eusocial. Therefore, Ceratina bees are equally important for studying the social evolution as other Xylocopinae bees such as Xylocopa or Allodapine bees or also Halictid bees (Hogendoorn and Velthuis 1999; Schwarz et al. 2007). It is interesting that the knowledge about natural history and social biology of Ceratina bees is less than in previously mentioned groups, although the collection of data is much easier.

Ceratina bees have been intensively studied mainly in temperate regions, where solitary species are common (Lawson et al. 2018; Groom and Rehan 2018), but data from the tropics showed that facultatively social species are most common in that region (Rehan et al. 2009, 2015; Groom and Rehan 2018). Cyprus, where we performed our research, lays in the subtropical zone and we showed that facultatively social species prevail there as well. Similarly, Japanese species which live in the warm-temperate or even the subtropical climate are also facultatively eusocial (Sakagami and Maeta 1977, 1987). Generally, the possibility for multiple broods per year is supposed to be an important factor for social nesting (Field et al. 2010; Purcell 2011). Ceratina bees have a long nesting season in Cyprus. Three species, C. mandibularis, C. parvula and C. cypricaca had active brood nests even in the late September, and in all these species, multifemale nests were documented. On the other hand, we have not observed any active brood nests in September in C. dallatoreana which had only solitary nests in our observation and in C. chrysomalla, which has a low proportion of social nests. The Japanese subtropical species C. okinawana in which multifemale nests occur has three peaks of reproduction per year (Okazaki 1987). Therefore, the season in subtropical climate seems to be sufficiently long for multiple broods per year.

4.2 Features of social nests

Number of individuals

Ceratina bees build linear nests; therefore females need to pass each other when one of them leaves or enters the nest. Linear nests do not allow multiple females to pass by each other easily; therefore the number of individuals per nest is constrained by nest architecture in Ceratina (Rehan et al. 2009). Two females in Ceratina nests perform an effective division of labor. One performs foraging, and the other performs guarding of the nest (Sakagami and Maeta 1984; Hogendoorn and Velthuis 1999; Rehan et al. 2010). However, more females probably cannot work effectively, because only one brood cell is provisioned at a time. Nest architecture of Ceratina nests is probably constrained by the increase in colony size in Ceratina bees more than in halictid bees, where multiple brood cells can be provisioned (Schwarz et al. 2007; Michener 2007), or in allodapine bees with progressive provisioning (Schwarz et al. 2007).

Social Ceratina nests usually contain only two females and can contain up to four females (Sakagami and Maeta 1984; Okazaki 1987; Rehan et al. 2010). Although two-female nests were the most common pattern in social nests in all social species in this study, we repeatedly detected nests with more individuals. We found even six females in a nest of C. parvula. Moreover, we found one nest with five females and four females frequently occurred in the nests of C. mandibularis. Therefore, these species can build unusually large societies in comparison to other Ceratina bees.

Males

In rare cases, we found adult males present in active brood nests and full brood nests. We observed male-female pair in C. mandibularis and C. cypariaca. Moreover, we found adult males in multifemale nests in C. cypriaca and C. bispinosa. These males can be accidental visitors and/or also male helpers which philopatrically stay in the nest from the previous nest clutch. We suppose that none of the studied species have regular biparental strategy, which is known in C. nigrolabiata (Mikát et al. 2019b). Also, male behavior is different from C. chalybea, where young males prevail in social nests (Mikát et al. 2021a). However, larger datasets, nest observation, and genetic tests are necessary for uncovering the role of males in these nests.

Nest productivity

Social nesting should be beneficial, if nesting productivity per individual increases. Surprisingly, empirical studies commonly show a decreasing productivity per individual (Michener 1964; Richards 2011; Prager 2014). Studies performed on Ceratina bees up to this date showed different results. Number of brood cells per female does not differ in C. japonica (Sakagami and Maeta 1984) but significantly decreases in social nests of C. australensis (Rehan et al. 2014; Dew et al. 2018), and C. chalybea (Mikát et al. 2021a). We found that social nests are more productive than solitary in three species with enough data, but the number of brood cells per female did not differ between solitary and multifemale nests. As per capita productivity is the same for both strategies, their coexistence is not surprising. However, the benefit of social nesting is probably higher than measured only by a number of brood cells provisioned per female, because social nests have smaller probability of nest failure (Smith et al. 2007). Therefore studies measuring the productivity primarily as a number of brood cells per female can underestimate the benefits of social nesting. Studies, which take also other factors such as brood or nest survival into account, showed commonly increasing per capita productivity in small societies (Hogendoorn and Zammit 2001). Social nests usually have a higher colony survival than solitary nests (Hogendoorn and Zammit 2001; Smith et al. 2007). In Ceratina bees, orphaned solitary nests are usually destroyed by natural enemies (Mikát et al. 2016), and social nests should have a lower probability of nest orphanage, although this phenomenon was not tested in Ceratina bees to this date.

Nest architecture

Social nests were generally longer than solitary nests, although the difference was significant only for C. parvula and C. mandibularis and not for C. cypriaca. As multifemale nests have higher productivity, it makes sense that they need more space for brood cells. There can be two reasons for this difference: higher probability of establishing social nests in larger burrows or enlarging and cleaning the burrow after the first brood clutch.

Very good nest structure feature is a distribution of empty cells in the nest. We found that empty cells are more often present in solitary nests than in social nests in C. mandibularis. Empty cells are considered to be a protection against natural enemies (Tepedino et al. 1979; Münster-Swendsen and Calabuig 2000) and more effectively guarded multifemale nests probably do not need this protection. However, we have not found a difference in empty cells proportion in C. cypriaca nor C. parvula. Empty cells are scarce in both species; therefore they probably rely on other types of nest protection.

Dominance hierarchies

In multifemale nests, usually multiple females had at least some ovarian development. Therefore, we can suppose that reproductive skew is incomplete and more females can participate in reproduction. Incomplete reproductive skew was documented for Japanese Ceratina from the subgenus Ceratinidia (Maeta and Sakagami 1995; Hogendoorn and Velthuis 1999) and it is generally common in small insect societies (Sherman et al. 1995). In contrast, in Xylocopa bees, a complete dominance of one female on reproduction was documented (Hogendoorn and Velthuis 1999). However, ovarian development itself is not a sufficient indicator of reproduction (Cini et al. 2013), and analysis of relatedness is necessary to uncover the reproductive dominance.

We found no relationship between the head width of female and the ovarian development. Therefore, we can suppose that the size is not important for establishing dominance hierarchies in these species. This is different from the situation in C. japonica, where the larger females usually have more developed ovaries (Sakagami and Maeta 1984). Our results show that reproductive dominance is probably related to age—females with higher wing wear also usually have more developed ovaries. Age-based and not size-based reproductive hierarchies are also known in Stenogastrinae wasps (Bridge and Field 2007). In Xylocopa bees, age is an important determinant of reproductive dominance, and an older female is usually dominant; however, too senescent females can be reproductive subordinates (Hogendoorn and Velthuis 1999). Size is important for establishing the dominance hierarchies in Xylocopa, when females are of the same generation (Hogendoorn and Velthuis 1999). As social nests are relatively scarce, our sample size is small for a deeper insight into reproductive hierarchies in our studied species. Later studies with a larger sample size are necessary for a better understanding of the reproductive division of labor in Ceratina bees.

4.3 Nesting biology of Ceratina in Cyprus

Nest guarding

Although Ceratina bees usually guard their nest until adulthood, facultative nest abandonment has been documented in two European Ceratina species (Mikát et al. 2016, 2019a). We found only a few full brood nests without a female in the most currently studied species. About 25% of full brood nests without female were found in C. chrysomalla and C. dallatoreana. However, we have not found a clear distinction in nest architecture between nests with mother and without mother in these species in contrast to the situation in C. chalybea and C. chalcites, where abandoned nests are closed by a filling plug that is much thicker than the usual nest partition (Mikát et al. 2016, 2019a). Therefore, we suppose that female always tries to guard her offspring until adulthood in all species here, and she does not perform facultative nest abandonments. However, further research of guarding strategy is necessary for the confirmation of obligate nest guarding for C. dallatoreana and C. chrysomalla.

Branched nests

Branched nests are common in Xylocopa bees (Gerling et al. 1983; Steen and Schwarz 2000). However, it has not been documented for Ceratina bees yet (Eardley and Daly 2007; Vickruck et al. 2011). Ceratina bees usually nest in relatively narrow twigs, where there is not enough space for a branched nest. However, C. cypriaca commonly uses broken Ferula stalks with about 2 cm in diameter where there can be enough space for a branched nest. We have documented only one branched nest yet. This situation can be either a mistake or a regular strategy of species. But this phenomenon deserves future research due to a possible influence of nesting architecture on social organization (Steen and Schwarz 2000).

5 Conclusion

We showed that facultative sociality prevails in Ceratina species in southern Europe (Cyprus). However, social nests are not common in facultatively social species, and their proportion was between 3 and 27% of nests, consistent with to what is known from previous studies about facultatively social Ceratina. However, we detected nests with 5 or 6 females, but not more than 4 females nesting together was recorded in Ceratina bees to this date. Social and solitary nesting are probably two similarly effective strategies, as per capita productivity is similar for both strategies. Usually, multiple females with developed ovaries are present in multifemale nests; therefore the reproductive division of labor is probably not complete. Reproductive dominance is probably related to the age, but not to the size of a female.

Table I Number of brood cells provisioned in active brood nests and full brood nests in different species. Only non-orphaned nests are included. Difference in number of brood cells between active brood nests and full brood nests was tested by Wilcoxon test
Full size table
Table II Number of nests with different social status. Only active brood nests and full brood nests were included. Solitary nests contained only one female, multifemale nests contained at least two females (and sometimes also males). Orphaned nests contained no female, rarely contain male or dead female. Male + female nests contained one male and one female. Dead + solitary nests contained one living and one dead female
Full size table
Table III Number of solitary and multifemale nests in active brood nests and full brood nests. Numbers of nests with other social status was not shown
Full size table
Table IV Comparison of solitary and social nests. SD standard deviation
Full size table