Nesting of Ceratina nigrolabiata, a biparental bee

Biparental care is very rare in insects, and it was well-documented in only one bee species to this date – Ceratina nigrolabiata. However, biparental care was only recently discovered in this species, and detailed description of natural history of this species is missing. Here, we describe the nesting cycle of C. nigrolabiata. Pairs of C. nigrolabiata are established before female starts offspring provisioning. After provisioning is finished (when youngest offspring reached larval stage), the male abandons the nest. Males which are present in nests where female already finished provisioning brood cells, are probably mainly temporary visitors. The female can perform long-time offspring guarding, but only 22% of completely provisioned nests are guarded by a female. Most nests (54%) are closed and abandoned, when provisioning is completed, and other (24%) are orphaned before provisioning is finished. Guarded nests have statistically higher number of brood cells provisioned than unguarded nests. Generally, C. nigrolabiata is unique among bees due to its biparental behavior, but it has also uncommon traits of nesting biology among Ceratina bees, e.g. fast offspring development in comparison with provisioning rate, and high proportion of nests which are closed and abandoned by mother.


Results
Phenology. Ceratina nigrolabiata excavate new nests mainly in May and June, however, some newly excavated nests were also recorded later in the season (Figs. 1, 2). Active brood nests (Table 1) occurred from half of June and appeared in high proportion through whole July. First full brood nests first occurred at the end of June, but the main peak of full brood nests was in July. Full brood nests were also frequent in August. Full-mature and mature brood nests occurred from the end of July, and they were very frequent through August. Other types of nests occurred mainly in the beginning and at the end of season. At the beginning of the season occurred mainly old hibernacula or adults of C. nigrolabiata visiting nests of other Ceratina. In the late phases of season occurred abandoned nests with only parasites and newly excavated burrows for hibernation.
Type of nest founding. We found two types of newly founded nests. Newly excavated nests, which were built by excavating pith from a twig. Discarded nests are the other type. These nests were built from previous nest of Ceratina (probably other C. nigrolabiata in most cases) by discarding a part of or all original offspring (Figs. S1 and S2). We observed nests of C. nigrolabiata, where nest partitions were destroyed and pollen from brood cells was placed on side of the nest. We suppose that original offspring were discarded out of the nest (and where egg is present at the top, but young adults already developed at the bottom of nest (f) guarded full brood nest-mother guards this nest (f) plugged full brood nest-nest is unguarded and closed by a thick filling plug (g) orphaned full brood nest-last brood cell partition is thin and above it is commonly pollen from incompletely provisioned brood cell (h) full-mature brood nest-this nest contains juveniles, young adults, and sometimes mother (i) mature brood nest-this nest contains young adults and sometimes mother. All these figures are hypothetical examples, they are not based on concrete dissected nests. www.nature.com/scientificreports/ on several occasions, we observed discarding of offspring out of the nest). Pollen provisions of the previous nest owner were usually moved to the sides of the nest (Fig. S1). From newly founded nests, 82.69% (86/104) were newly excavated and 17.30% (18/104) were discarded nests. When we counted only nests founded after half of June, the proportion of discarded nests was 22.78% (18/79     in nests, where female was also present, but only 1 day on average (N = 12, SD = 0) in nests where only male was present. In full-mature brood nests, male stayed 3.33 days on average (N = 6, SD = 2.5, range 1-8) at the time of nest dissection. Sample size is too small for testing the difference between nests where a female was present and absent, but we observed both males which stayed in a nest one day and males which stayed multiple days.
Paternity of nests with small number of offspring. Guarding male was usually not the father of offspring in young provisioned nests. Guarding male was the father of 6.25% (10/160) of all offspring and 9.9% (10/101) of female offspring in nests with 1-3 offspring (N = 70 nests). No offspring was fathered by guarding male in nests with only one offspring (N = 17). The proportion of offspring guarded by it´s own father was 6.2% (2/32) in nests with two offspring and 7.2% (8/111) in nests with three offspring.

Structure of full brood nest.
Full brood nests contained on average 7.59 brood cells (range 1-21, SD = 3.76, N = 566). Empty cells were relatively scarce, but usually present. There were 1.33 (range 0-8, SD = 1.33, N = 530) empty cells per nest on average. Therefore, brood cells were usually adjacent. However, when an empty cell was present, it was usually much longer than a brood cell (Fig. 3). Length of nest was 15.12 cm (N = 670, SD = 4.08) on average and entrance burrow was 3.63 cm long (N = 657, SD = 3.63) on average. We distinguished three types of full brood nests: (1) guarded nests, (2) plugged nests, (3) orphaned nests. In guarded nests, an old female was present. The Last brood cell was always closed by nest partition. Other two nest types, plugged or orphaned nests, were without presence of an old female. Plugged nests had the last brood cell closed by a filling plug. Filling plug was much thicker than the regular nest partition (Fig. 3) Filling plug was 1.41 cm long (N = 307, SD = 0.87, range 0.2-8.0) on average. Moreover, plugged nests had usually modified nest entrance. All pith between nest entrance and a filling plug was excavated. In orphaned nests, the last brood cell was closed by a regular partition, which was not thicker than regular brood cell partition. Sometimes, the last brood cell was partially provisioned by pollen and was not closed by brood cell partition. This type of brood cell did not contain living offspring.
There was a distinct difference in the proportion of full brood nest strategies through the season. The proportion of plugged nests was highest at the beginning of full brood nest season (beginning of July) and later decreased. On the other hand, the proportion of guarded and orphaned full brood nests increased from a beginning of July to August (Fig. 4). The proportion of full brood nest types significantly differed between different periods in season (Chi-square test, Chi = 116.87, df = 10, p < 2.2e−16).
Parasites. The most common nest parasites were Ichneumonidae and Gasteruption. Both destroyed multiple brood cells and commonly destroyed a large part of a nest. We found an ichneumonid parasite in 6.81% (125/1836) of nests and Gasteruption in 2.83% (52/1836) of nests. In 38 cases, we were unable to determine if the parasite is an ichneumonid or Gasteruption. When we suppose the same proportion of both parasites in determined and undetermined larvae, we can assume that 8.26% of nests were parasitized by Ichneumonidae and 3.44% by Gasteruption. Usually, there was only one larva of these parasites per one nest. We observed 8 cases of two Ichneumonidae larvae in one nest, 2 cases of two Gasteruption larvae in one nest and 2 nests where Gasteruption and ichneumonid larvae were present together.
Proportions of attacked nests differed between nesting phases. No new founding nests were attacked, as there is no food for the parasite. In active brood nests, only 2.74% were parasitized by Ichneumonidae and only 1.43% by Gasteruption. In full brood nests, 11.68% were parasitized by Ichneumonidae and 4.83% by Gasteruption. The complete number of parasitized nests is summarized in Table 4.

Discussion
Role of males. The male-female pair is established in C. nigrolabiata before provisioning of brood cells starts. During the period of brood cell provisioning, a pair is present in almost all nests. However, the pair is not stable, and the male may sometimes vanish and is later replaced by another male 32    www.nature.com/scientificreports/ than males. A similar situation occurs in later nest stages. Therefore, males have no important role in brood care after brood cell provisioning is finished and a female stays with her offspring in only a minority of nests. Biparental care is an uncommon type of parental care in insects 12,15 and from all bees it is confirmed only in C. nigrolabiata 32 . There is an extensive division of labor between males and females in C. nigrolabiata. Female does all nest provisioning, but male participates in nest guarding 32 . From other Hymenoptera, biparentality is well documented in several species of the genus Trypoxylon 30,31,38 . Males in Hymenoptera are usually short-lived and die shortly after mating 20,39 , therefore, there is a little possibility for performing any care. However, some male participation in care is known also in other Hymenoptera species 26,40 .
Males of C. nigrolabiata were present in nests before provisioning started. They were present in newly excavated and discarded nests. New founded nests with male-female pair were more common than nests with only female. Male can help female with nest excavation by throwing filling from a nest or by discarding offspring of previous nest owner. Therefore, males have a partial role with nest building, similarly with males in crabronid wasps from genus Trypoxylon, where males help with smoothing of mud using their mandibles 29 .
Although a male is commonly present in the nest before provisioning starts, he is usually not the father of offspring which female laid immediately after she starts provisioning. We found out that the guard male was the father of only 6.25% offspring in nests with 1-3 provisioned brood cells. This proportion is even smaller than the average proportion of offspring guarded by own father which was 10% 32 . Therefore, it is evident that the female mates before provisioning season and the male comes to nest primarily as a stepfather. This situation is in contradiction to other biparental insects, where biparentality is based on monogamy and therefore high relatedness between father and offspring 12 . A minority of offspring (45%) is cared for by father also in the passalid beetle Odontotaenius disjunctus 41 .
Our results show that males are present in nests where receptive females are also present. They are often in newly founded nest and in almost all active brood nests. However, they scarcely occurred in full brood nests or mature brood nests. In full brood nests, a male was often in nests where a female was also present. Moreover, when the female is removed from active brood nests, the guarding male usually disappears after few days 32 . Therefore, the main male motivation is mate-guarding behavior, not direct offspring care. Males in full brood nests and mature brood nests stay only one or a few days. Therefore, they cannot be fathers of any offspring in the nest. We suppose that two motivations for male presence are possible: (a) mating with newly emerging young females and (b) staying overnight in the burrow in case of single males.
Although male primary motivation is mate-guarding, our previous study shows that male is beneficial for nest productivity 32 . We suppose that presence of a male in the nest is useful as protection of nest when a female is on a foraging trip. In active brood nests is male present in vast majority of nests and when a male is not present, female foraging activity strongly decreases 32 .
Generally, a behavior of males and females of C. nigrolabiata is similar to biparental species of genus Trypoxylon. In Trypoxylon, females also perform all nest provisioning and males stay at the nest entrance and protect the nest against natural enemies 29,38 . Biparental care is supposed to be a by-product of mate-guarding in both groups 31,32 . Some differences between Trypoxylon and Ceratina exist. In Trypoxylon, males stay in the nest entrance head out 38,42 , which allows them to guard more actively than C. nigrolabiata males, which block nest entrance by metasoma. Moreover, Trypoxylon males usually spent the night outside the nest 29,38 , but males of C. nigrolabiata do not leave the nest at night. Generally, we can consider C. nigrolabiata and Trypoxylon as taxa which convergently developed very similar biparental behavior. Moreover, it is possible that similar behavior occurs also in colletid bee Leiproctus muelleri. In this species, males perform nest guarding when female provisions nest 43 . However, more detailed research is necessary for evaluation of the male role in this species. Behavior of males is different in all other hymenopteran groups, in which they assist with caring for offspring. In small eusocial colonies of Microstigmus nigrophalmus, males help with nest protection, however there are more males in the nest and they don't sit in the entrance, but patrol across the whole nest 26 . Male participation on care was detected in some polistine wasps or bumblebees and stingless bees, but the role of males is only small and males help with thermoregulation or food processing 24,25,27,44 . Macrocephalic males were documented in Lasioglossum (Chilalictus) erytrurum, which can guard nest against ants 45 . However, these males were observed in the late phase of nesting, when no brood was produced. Therefore, they probably guard their siblings, not offspring. We suppose that biparental care in Hymenoptera can emerge more easily in species which build linear nests with easy defensible nest entrance, where a male can perform nest guarding. One guarding male is less effective in other types of nest architecture. Male participation on care which emerged in other hymenopteran groups (e.g. eusocial Polistes wasps or Bombus, macrocephalic males in bees) is not based on pair living. Probably different selection pressures favor its emergence than pressures for typical biparental care which occurs in Ceratina nigrolabiata and Trypoxylon.
Alternative nesting strategies. Parental care is costly and reduces future reproduction 1,10,46 . Therefore, animals optimize time when they leave their offspring 46,47 . Most non-eusocial nest-making Hymenoptera abandon the nest after provisioning is finished 20,48,49 , although guarding of nest can substantially increase offspring survival 35,45 . However, guarding of the nest by the female until offspring adulthood is typical for Ceratina bees 34,35 .
Our results show that C. nigrolabiata has alternative nesting strategies. Some females are trying to guard the nest until the adulthood of offspring. However, most females plug nests by a filling plug and abandon it. This facultative behavior was already documented in C. chalybea and C. chalcites 35,37 . We suppose that females, which abandoned their nest, build a second nest elsewhere. We do not have direct evidence for this statement, but we found newly founded nests and active brood nests also in late phases of the nesting season (Fig. 2). Moreover, almost all full brood nests were plugged in early phases of nesting season, but guarded nests prevailed in late phases of nesting season. Therefore, we suppose that females usually abandon their early nest(s) and guard their www.nature.com/scientificreports/ last nest. Females can probably abandon their nest, when there is enough time for second nesting. It corresponds with the semelparity hypothesis. It means that opportunities for reproduction can reduce the extent of parental care 50,51 . Abandonment of larger brood by mother is generally less probable than abandonment of smaller brood 2,52 . We found out that guarded nests have significantly higher number of brood cells provisioned than abandoned nests. However, we have not detected direct effect of guarding on offspring survival. There was no difference in the proportion of nests attacked by an ichneumonid or Gasteruption parasite between guarded and abandoned nests. The proportion of dead brood cells differed between nest types, but was lower only for orphaned and not abandoned nests in comparison with guarded nests. However, the most important cause of brood destruction in case of female removal in Ceratina bees is usurpation by other Ceratina or the nest destruction by ants 32,35 . This type of attack leads to the destruction of whole or a significant part of nests, but we were unable to detect such effect by a simple comparison of different nest types. Therefore, long-term observations of nest mortality are necessary for comparison of the success of guarding and abandoning strategies.
Guarded and plugged nests differ in the number of brood cells provisioned and length of nest entrance, though the overall difference in nest architecture was small between nest types. In C. chalybea and mostly also in C. chalcites, the last brood cell was open in guarded nests 32,35 . However, the last brood cell is closed in both nest types in C. nigrolabiata. Moreover, last nest partition can be enlarged to filling plug also in some guarded nests (Fig. 3).
Although most of nests without mother are voluntarily abandoned, we detected high proportion of nests (22%) which seems to be orphaned. This is an important difference from C. chalybea and C. chalcites, where orphaned full brood nests are extremely rare or completely missing 35,37 . Natural enemies. Ceratina bees are attacked by a wide spectrum of natural enemies. However, the influence of parasitism is usually low due to effective nest protection and short time of larval development 53 . The most common parasites, which we observed, were ichneuomids and Gasteruption. Both parasites have a similar effect on nests. Their predacious larvae are much larger than Ceratina, and they destroy several brood cells (Fig. S3). The number of broods destructed by one ichneuomonid or one Gasteruption is probably about four, but it is difficult to count them as partitions are damaged.
We suppose that the most relevant stage for assessing parasitation is full brood nests. Earlier nest stages had not sufficient time to be parasited. On the other hand, both parasites and Ceratina offspring can emigrate from later nest stages, thus full-mature brood nests and mature brood nests aren't suitable for assessing parasitation. As about 12% of full brood nests were parasitized by Ichneumonidae and 5% by Gasteruption, we think that these parasites cause substantial brood loss in this species. On the other hand, other brood parasites were very rare and they probably do not affect C. nigrolabiata population substantially.
Nest usurpation plays an important role in C. nigrolabiata strategy. From new founded nests, 18.2% were established by usurpation. Moreover, removing experiments show that usurpation by other Ceratina bee is the most important reason of failure of the nest with removed female 32 , and most of these usurpers are conspecific individuals of C. nigrolabiata. Therefore, interspecific competition plays apparently important role in C. nigrolabiata. However, it is a question, why some females frequently abandon nests. Frequency of unguarded nests is even larger than in related species C. chalybea and C. chalcites 35,37 . In plugged nests is the nest entrance usually excavated, and therefore its usurpation by other Ceratina is prevented. It is impossible to guard the nest effectively and therefore plugged nests are probably unattractive for nest usurpation.

Rate of development.
Ceratina nigrolabiata have excessively fast development in comparison to the duration of the provisioning period of the nest. Therefore, the largest active brood nests contain already adult offspring at the bottom. Active brood nests with adults at the bottom contained on average more offspring than full brood nests. Moreover, nest with the largest number of brood cells provisioned (23) were not full brood nest but active brood nest with adults at the bottom.
High rate of offspring development leads to less risk of nest abandonment by mother. Adult offspring crawl through nest partitions to the top. They can protect immature siblings against potential intruders soon after mother emigration.
High rate of offspring development complicates determining the average number of offspring in complete nests. The reason is that larger nests are in the stage of full brood nest for a shorter time, which is the only stage when counting of total number brood cells provisioned is possible. When offspring crawl though uppermost brood cell partition, they can emigrate from natal nest. Therefore, average number of brood cells provisioned can be underestimated due to lower probability of detection of a large nest. Moreover, the proportion of guarded nests can be also underestimated, because these nests are larger (and therefore less detectable) on average than plugged or abandoned nests. General design. We dissected nests from artificial nesting opportunities. Some of these nests were used also for other experiments (partially published in 32 ), but here we present different aspects of C. nigrolabiata biology than in our previous paper. www.nature.com/scientificreports/ Preparation of nesting opportunities. We studied C. nigrolabiata nests from artificial nesting opportunities. We used twigs of Solidago canadesnis, Echinops spareocephalus, Helianthus tuberosus, and Tanacetum vulgare. We cut twigs to 30-50 cm long fragments. Twenty of these fragments were tied together into one sheaf. Each sheaf was fixed by a bamboo rod to ground. The sheaves were installed before nesting season (April or early May). We established about 1000 sheaves, which corresponds to 20,000 nesting opportunities each year. Therefore, we established around 120,000 nesting opportunities during the whole research.

Methods
Nest dissection. Nests were collected throughout nesting season from May to September. In total we collected 19 nests in May, 149 in June, 1413 in July, 244 in August, and 11 in September. We dissected most nests in July, since the most important nest stages occur during this part of nesting season. Nests were collected in the evening (after 18 h CEST) to ensure that all inhabitants were present inside the nest. Nests were stored in a refrigerator between the time of collection and dissection. Each nest was open by a knife or clippers. Following parameters were noted for each nest: length of the nest, length of nest entrance, number and stage of immature individuals, number and sex of adult individuals, presence of parasites. Position in a nest was noted for each individual. We also noted the presence of nest partitions, which separated brood cells. We specifically noted presence of a filling plug (enlarged the last brood cell partition, which is usually about 1 cm long).
Nest stage classification. We classified nests into categories. Earliest occurred new founding nets (Fig. 1, Table 1), which contained burrows with only adult individual(s) and no pollen ball or provisioned brood cells. We divided such nests to two sub-categories: newly excavated nests, which were newly established, and discarded nests, which were established in twigs that already housed another Ceratina nest. In other words, the nest was usurped. Active brood nests contained a pollen ball in the outermost currently provisioned brood cell or an egg in the outermost closed brood cell. Moreover, these nests were not closed by a filling plug. Therefore, active brood nest is a nest, where provisioning of new brood cells is present at the time of nest dissection. Full brood nests contained a larva or pupa in the outermost brood cell, and the partition of the outermost brood cell was still undisturbed. If young adults were present in the nest, they did not crawl through the outermost partition. Full brood nests are nests, where provisioning is already finished, but young adults still did not disturb nest partitions. Therefore, nests at this stage are the most relevant for assessing nest structure. Only these nests are relevant for counting brood cells, because active brood nests are incomplete and young adults can emigrate from full-mature or mature brood nests. Full-mature brood nests had disturbed the outermost brood cell partition (mature adults probably crawled out through this partition) and contained at least one immature offspring. Mature brood nests contained only mature offspring and no juveniles. We analyzed 1,836 nests of C. nigrolabiata: 86 nests were newly excavated, 622 were active brood nests (460 of them were already used for same analyses published in 32 ), 672 were full brood nests, 177 were full-mature brood nests, 85 were mature brood nests, 18 were discarded nests, and 176 nests were impossible to classify as any standard category.
Nests, which were impossible to classify into a standard category, were old burrows used to stay overnight or for hibernation (N = 73). These burrows were excavated by a young adult at the end of season or commonly contained evidence of activity of other arthropods (brood cells of other Hymenoptera, spider net) at the bottom. Other nests were evidently build by C. nigrolabiata (N = 83), but have not contained any live adult or juvenile C. nigrolabiata individual. They contained excrements, dead offspring or only parasites. Another nests (N = 14) contained living C. nigrolabiata offspring, but they were distinctly damaged by natural enemies and therefore it was impossible to determine their stage. Last non-standard type of nests were nests which contained only brood cell partitions built by C. nigrolabiata, but without living or dead offspring (N = 6). These nests resembled full brood nests, but contained only empty cells (no provisioned brood cell was present).
Analyses. Phenology. We calculated the proportion of types of nests through different parts of the nesting season. Nests dissected in September (N = 11) were excluded from analysis, due to the small number of nests dissected in this period. Therefore, we included 1825 nests. We calculated the proportion of each nest type in each month.
Type of nest founding. We distinguished two types of newly founded nests (nests where provisioning did not begin): newly excavated nests which were built as a new burrow, and discarded nests which originated from a previous Ceratina nest with its original content discarded by new owners (Fig. S1). We calculated the proportion of both types of nest founding. Moreover, we checked active brood nests and noted evidence of previous discarding (pollen attached on the sides of the nest, Fig. S1).
Presence of parents. We calculated proportion of nests where male, female, male-female pair or no adult of parental generation was present. We also counted the proportion of nests where more than one adult of one sex of parental generation was present (In full-mature and mature brood nests were also present young adults. Old and young adults can be distinguished by wing wear). We tested differences in the proportion of nests guarded by different sex by chi-square test. In new founded nests, we tested difference in nest length between nests guarded by a male, a female, or a pair. Firstly, we tested difference by Anova and later we used Tukey HSD tests for pair comparisons. All statistical analyses were performed in R software 54 .
Duration of guarding of current male. We measured how long was a nest guarded by a male which was present at the time of nest dissection. We performed this analysis in years 2013-2016. For this experiment, we used all www.nature.com/scientificreports/ nests in one sector of studied locality. We daily checked the presence and identity of guarding male. We observed nests during the day (between 9 and 17 CEST), when observation of male is easiest, because he stays near nest entrance. We checked each nest once each day. When we found an unmarked male, we marked him with an oil dye. When we found a marked male, we noted his color. Therefore, we were able to determine how long the male is present in the nest. We performed marking of nests through whole provisioning season, from about mid-June to the end of July. We randomly selected nests for nest dissection through this period. Nest stage was determined at the time of dissection. We examined 302 active brood nests, 30 full brood nests, and 6 full-mature brood nests. In this analysis, we included only nests which were observed for at least ten days, though most nests were observed for a longer time. Active brood nests were observed on average for 20.9 days (range 10-41). Full brood nests were observed on average for 25.1 days (range 13-37). Full-mature brood nests were observed on average for 28.2 days (range . In 27% (83/302) of active brood nests and 3.3% (1/30) of full brood nests, the same male was present throughout the whole observation period. Therefore, the duration of actual male presence is underestimated in active brood nests and probably correctly calculated in full brood nests and mature brood nests. For a detailed description of this method, see 32 . For full brood nests, we tested the duration of presence of currently guarding male between nests where a male was alone and where a male was in pair with a female. We used quasi-Poisson GLM with year and duration of observation period (in days) as covariable. Quasi-Poisson model was used because overdispersion was present. Analysis was performed in R software 54 .
Paternity in small nests. For this analysis, a subset of nests analyzed in previous study was used 32 . Procedures of DNA isolation, microsatellite genotyping, and paternity analysis are described in 32 . Nests were selected for analysis according to these characteristics: (1) Nests had between one to three provisioned cells with offspring, (2) guarding parent pair was present, (3) all offspring in the nest was offspring of guarding female. 17 nests with one offspring, 16 nests with two offspring, and 37 with three offspring were included in this analysis. In these nests, we tested paternity of the guarding male.
Full brood nest structure. We calculated the number of provisioned cells, empty cells, and living offspring for full brood nests and active brood nests. The number of provisioned cells was possible to calculate only in nests which were not influenced by an ichneumonid or Gasteruption parasite, or only slightly influenced. These parasites destroyed multiple brood cells, and it was impossible to determine the precise number of brood cells destroyed. The number of empty cells was possible to be calculated in nests which were not parasited by Gasteruption or Ichneumonidae, and if any adult offspring destroyed partitions in the bottom part of the nest. When adult offspring destroyed brood cell partition of at least one brood cell, it was impossible to determine if an empty cell were destroyed or not.
Comparison of different full brood nest types. Full brood nests were classified in three categories: guarded, plugged, and orphaned (Table 1, Fig. 1). Guarded full brood nests contained an old adult female in the nest entrance. Last brood cell was always closed. Plugged nests did not contain female at nest entrance and were closed by a filling plug (nest partition on average 1.41 cm thick, much more than regular brood cell partition). Last brood cell was always closed. Orphaned nests did not contain female at nest entrance and the last brood cell partition was of the same thickness as the other partitions in that nest. Sometimes, the last brood cell was opened, without living offspring and only partially provisioned by pollen in orphaned nests. Adult males can be present in nest entrance of all types of these nests. We did not use the presence of male as a factor for nest classification. We tested differences in nest features between three nest categories using Anova tests as a covariable. We used equation with interaction (dependent variable ~ year * nest type) or without interaction (dependent variable ~ year + nest type). The decision to include the interaction or not was based on model AIC. Later, we used TukeyHSD pair comparisons when significant Anova results were found. The statistical analyses were performed in R software 3.6.1 54 .
Parasitism. We analyzed the presence and number of natural enemies at the time of nest dissection. We were unable to distinguish between early stages of ichneumonid and Gasteruption larvae. Therefore, we inserted these larvae in Eppendorf tubes and tried to rear them at least to prepupal stage when these two parasites are easy to distinguish. Some larvae died before this stage (N = 38). For assessment of total parasitation rate, we divided these cases proportionally between ichneumonids and Gasteruption. We tested the association between presence of parasites which attack multiple brood cells (ichneumonids and Gasteruption) and full brood nest type (guarded vs. plugged vs. orphaned) by chi-square test. Moreover, we tested differences in proportion of brood cells with dead offspring between full brood nests types. For this analysis, we excluded nests parasitised by an ichneumonid or Gasteruption because the number of damaged brood cells was not easy to count. We tested this difference using binomial generalized linear model because the proportions of dead brood cells have binomial distribution. We used year as covariable. We used equation with interaction (dependent variable ~ year * nest type) or without interaction (dependent variable ~ year + nest type). The decision to include the interaction or not was based on model AIC. The statistical analyses were performed in R software 3.6.1 54 .
Analysis of developmental stage diversity in active brood nest. For active brood nests, we noted the stage of offspring in the innermost (= oldest) brood cell. We calculated the proportion of nests with adult, pupa, larva, and egg in the innermost brood cell. Moreover, we calculated the average number of brood cells for active brood nests with each offspring stage separately. www.nature.com/scientificreports/

Data availability
All relevant data are attached in XLS file as supplementary material.