Apidologie (2023) 54:51 Original article © The Author(s), 2023 https://doi.org/10.1007/s13592-023-01030-y The effect of forager loss on honeybee workers temporal polyethism and social network structure Elias Beck1 , Tom Wenseleers1 , and Ricardo Caliari Oliveira2 1 Laboratory of Socioecology and Social Evolution, Department of Biology, KU Leuven, Naamsestraat 59, 3000 Leuven, Belgium 2 Departament de Biologia Animal, de Biologia Vegetal i d’Ecologia - Universitat Autònoma de Barcelona, 08193 Bellaterra, Barcelona, Spain Received 13 October 2022 – Revised 25 August 2023 – Accepted 21 September 2023 Abstract – The western honeybee Apis mellifera is perhaps the best-known example of an advanced eusocial species displaying temporal polyethism, a process in which workers perform different tasks in the colony as they age. Previous studies demonstrated that this temporal division of labour is not only regulated by the age of the bees but also by the proportion of workers performing the required tasks in the colony. As we progress through the Anthropocene, the Intergovernmental Panel on Climate Change predicted that a significant increase in both the frequency and intensity of severe weather events can be expected in the coming decades. Here, we performed a controlled interrupted time series experiment with the objective of quantifying the possible effects that these weather events might have in the honeybee colony dynamics. By simulating a significant loss of foragers in the colony, we observed that honeybee workers quickly replaced the missing foragers by accelerating their transition to a subsequent task and, in some cases, completely skipped several of the natural task transitions with respect to their age. In addition, we analysed how the colony social network structure is affected by the sudden loss of foragers. Indeed, our data show that honeybee colonies increased their network cohesion with workers having a higher number of interactions as well as becoming more closely connected to one another soon after the dis- turbance. Overall, our data shows that even when faced with a substantial perturbation, honeybees can respond swiftly in order to maintain colony homeostasis and likely increase their resilience against future perturbations. Apis mellifera / age polyethism / task allocation / social network / climate change 1. INTRODUCTION overlapping adult generations and the presence of reproductive and non-reproductive castes (Wilson Social behaviour is ubiquitous in nature, with var- 1971). The last trait is also used to distinguish ying degrees of social organization being observed between primitive and advanced eusociality with across the animal kingdom and beyond (Székely respect to whether there are noticeable morpho- et al. 2010). The most advanced level of social logical differences between the female castes or organization, namely eusociality, was first described not (Michener 2007). The best known example of in the Hymenoptera but later also found to occur an advanced eusocial species is perhaps the west- in several other taxa (Gadagkar 1987; Grimaldi ern honeybee Apis mellifera, which presents clear and Engel 2005). Eusocial, or “truly” social, species Vol.:(0123456789) behavioural and morphological differences between are characterized by having cooperative broodcare, the queen that monopolizes reproduction and the workers that specialize in colony maintenance tasks such as feeding the young and defending the colony Corresponding author: R. C. Oliveira, ricardo.oliveira@uab.cat Manuscript editor: Mathieu Lihoreau (Michener 1969; Peso et al. 2016). 13 51 Page 2 of 13 E. Beck et al. This division of labour is a hallmark of (IPCC), severe precipitation events are likely to eusociality and is not only characterized by have increased in their occurrence and intensity the reproductive division of labour between and will continue to do so in the future (IPCC queen and workers but also among workers 2014). Extreme rainfall and storm conditions can themselves. Indeed, rather than specializing in be detrimental for honeybees, particularly when fixed morphological castes, honeybee workers workers are out foraging. Rainfall can decrease perform different tasks as they age in a temporal their body temperature leading to a chill coma- division of labour pattern called age polyethism tose status where the honeybee metabolic activity (Robinson 1992). Almost immediately after ceases to a minimum (De Joy 1998). In addition, emerging from the brood rearing cells, they rainfall can have a significant detrimental impact begin cleaning the cells previously occupied on broodcare behaviour (Schmickl et al. 2003; by brood, transitioning to caring for the young, Schmickl and Crailsheim 2002). Strong winds storing food resources, defending the nest and, may also hinder foragers from returning to the finally, to foraging (Johnson 2008; Seeley and hive, increasing the probability that they do not Kolmes 1991). Nevertheless, age polyethism survive these events (Neov et al. 2019). is not fixed and workers may in fact shift tasks Aiming to quantify the possible effects that according to the colony needs (Bernadou et al. these weather events might have on the honeybee 2015; Naug and Gadagkar 1998; Woyciechowski colony dynamics, we performed an interrupted and Moroń 2009). That is, not all bees perform all time series experiment where we simulated the tasks for the same amount of time. For example, effect of an extreme weather event on the colony some workers in poor health will make a faster and tested how bees may adapt their task alloca- transition to the riskier foraging task (Lecocq tion in response to a significant loss of foragers. et al. 2016; Natsopoulou et al. 2016; Tofilski Whether colonies are resilient and able to swiftly 2009). The overall colony condition might adapt to adverse conditions leading to a signifi- also influence the temporal division of labour cant impact on colony homeostasis might have with some honeybee colonies being observed direct implications on the maintenance of stable to adapt their social organization according to populations in future climate scenarios. More nutritional resources requirements or to specific specifically, we tested whether the time taken for environmental changes (Schmickl and Crailsheim young workers to transition to a subsequent task 2002, 2004). This flexibility in the age polyethism is dependent on the proportion of foragers pre- enables the workers, and ultimately the entire sent in the colony, accounting for the effects on colony, to adjust their social organization in order colony social stability. In addition, we mapped to optimally exploit resources as well as to deal the social interactions of a subset of workers with both biotic and abiotic adversities (Robinson from all age cohorts during an entire generation 1992; Seeley 1995). and compared several social network properties The number of wild populations of honeybees before and after the experimental manipulation. in both Europe and the USA has been decreasing rapidly in the last decades, up to a point where almost no feral honeybees are left (Jaffé et al. 2. MATERIAL AND METHODS 2010; Panziera et al. 2022). Among the most important drivers for this ongoing pollinator crisis 2.1. Experimental setup are changes in land use, habitat fragmentation, the use of pesticides, pollution, invasive species, The experiments were conducted in the api- pathogens and climate change (Cox-Foster ary of the Laboratory of Socioecology and et al. 2007; Hendrickx et al. 2007; Kevan et al. Social Evolution at KU Leuven (50°52′27.6″ N, 1997; Potts et al. 2010; Stout and Morales 4°41′44.3″ E) in three one-sided experimental 2009; Winfree et al. 2009). According to the observation hives (hives A, B and C) of A. mel- Intergovernmental Panel on Climate Change lifera carnica honeybees consisting of a single 13 The effect of forager loss on honeybee temporal polyethism Page 3 of 13 51 standard frame covered by a transparent sheet of rainfall or storm. During the manipulation, 50% plexiglass. Two weeks prior to the beginning of of the foragers were removed (from a total of 15 the experiments, empty bee frames were added in hive A, 16 in hive B and 23 in hive C) by add- to eight beehives in the laboratory apiary which ing a sampling box at the entrance of the hives served as stock colonies for the observation hives. during the evening and collecting, counting and Observation hives were provisioned with water removing the foragers in the morning of the next and sugar paste ad libitum during the first days day. The remaining 50% of the foragers that were and presented a regular colony development until not removed were reintroduced to their respec- the experiments were performed. Once foraging tive observation hives. activity was established, we ceased feeding the experimental hives. We then selected three of these colonies and collected the frames together 2.2. Behavioural observations with the queen and a subset of about 200 workers and placed them onto the observation hives. In The recordings were analysed with VLC addition, we collected frames containing capped media player (VideoLAN Organization), brood cells from the same stock colonies and kept whereby five tagged bees were randomly them in an incubator at constant temperature (36 selected from all age cohorts that were present °C) and humidity (50%). For the next 3 weeks, at each recording day. The individual bees were bees emerging in the incubator were individually closely followed for 3 min after which its specific marked with numbered opalite bee tags at every task in the hive was determined. The specific 2 days. During each session, we marked 20 bees tasks consisted of the exact behaviour the bees with a different colour and introduced them into were performing during the observation. This their corresponding observation hive in order to behaviour included (1) nest maintenance work, replace the majority of bees in the hives by indi- (2) performing broodcare, (3) working in food viduals of known age (total of 10 sessions and storage, (4) guarding behaviour and (5) forag- 200 bees added per observation hive). ing. Workers were considered as foragers when Once all observation hives contained marked they were observed leaving and entering the hive bees of most age groups, we recorded the hives entrance during observation sessions. Workers at every 2 days with a Panasonic camcorder that did not fit any of the previous categories (HC-VX870) in 4K resolution for 15 min for 3 but still interact with nestmates were labelled as weeks. The recording started 2 weeks after we “idle” for the social network analysis. Aside from began introducing tagged newly emerged bees the tasks of the focal bees, we also noted the and persisted for 2 weeks. Hence, newly emerged individual ids of the bees in which the focal bee tagged bees were still introduced 1 week after interacted with as well as the type of interaction the beginning of the recording sessions. The (i.e. antennation, trophallaxis or interactions with experimental removal of foragers was performed the queen). The behavioural observations were 1 week after the start of the recordings in order not performed blindly to colony identity. Nev- to have a baseline “control period” before the ertheless, they followed a precise behavioural manipulation with four recording sessions and catalogue in order to clearly identify the tasks an “experimental period” with a recording on performed by workers. To determine the tasks of the day of the manipulation plus four subsequent the bees from the same age group which were not sessions. Therefore, we performed nine behav- directly observed, we analysed the task that was ioural data observations (i.e. four before treat- most common among the bees that were directly ment, on the day of the treatment, and four after observed within the same age cohort. When no the experimental manipulation). The interrupted specific task was single outed as the most com- time series experiment was designed to simulate mon, we then assigned a task from the equally an abrupt and significant loss of foragers due to possible tasks in that age group randomly. This an extreme weather event such as a sudden heavy particular dataset of inferred tasks was utilized 13 51 Page 4 of 13 E. Beck et al. exclusively for the purpose of completing the a subsequent task) as dependent variable with social network analysis. In addition, the entrance an interaction term between the timing of treat- of each observation hive was observed for 6 min ment (before or after) and the number of days to account for the bees performing extranidal between transitions. Finally, Tukey post-hoc tests activities such as guarding and foraging, to con- were performed to check for differences among firm whether the bees were indeed performing all possible pairwise transitions. extranidal tasks or were just at the hive entrance by chance. 2.4. Social network analyses 2.3. Statistical analyses We performed a social network analysis with the program UCINET version 6.738 (Borgatti All statistical analyses were carried out et al. 2002) to study the effects of treatment on using the R software version 4.0.2 (R Core the social network structures of the colonies. For Team 2020). In order to test the effects of the each observation hive, we created two datasets experimental manipulation in the age polyethism and subsequent social networks with one con- dynamics, particularly on the onset of foraging taining the tasks and one containing the inter- behaviour, we ran a multinomial model using the actions of the observed bees 4 days before and function multinom from the package nnet version 4 days after the treatment (see “Behavioural 7.3–14. The different tasks recorded in the data- observations” section for details). The interac- set were first grouped by similar behaviours into tion dataset contained both actors of each inter- five general task groups i.e. nest maintenance, action and the type of interaction for each day in broodcare, food storage, guard or forager. The each hive. This dataset was loaded in UCINET task groups were then coded as the response before constructing the network. After the crea- variable and an interaction term containing the tion of the network, the task dataset was added timing of treatment (before or after manipula- to the network. This task dataset contained the tion) and the age of the bees was coded as a fixed specific task of each observed bee per day of co-factor. Hive id was also added as a second each hive and functioned as the attribute table fixed factor. This was followed by an analysis corresponding to all the interacting bees in the of variance (ANOVA) to test the overall effects network. We then calculated network metrics of hive, treatment and age on the proportion of such as the diameter, density, global cluster- task allocation. Tukey post-hoc pairwise tests ing coefficient, average interaction rate, K-core were performed to check pairwise treatment index and Wiener index for all observation days effects within each specific task groups (pack- within each hive. The averages of the individ- age emmeans version 1.5.1). In addition, we ual metrics were calculated per hive before and tested if the experimental manipulation affected after treatment. These measures are indicators the time taken for bees to switch to a subsequent of the structure and connectedness and give a task. To this end, we first calculated the aver- good indication of the overall cohesion present age age of the bees performing each task and in a given network (Grandjean and Jacomy 2019; then subtracted from the following possible Tabassum et al. 2018). Next, a primary compo- task progression and age cohort according to nent analysis (PCA) was performed with the the literature and our own data (Johnson 2008; metrics of each social network before and after Seeley 1995; Seeley and Kolmes 1991). Chrono- treatment, and the primary components extracted logical transitions from one task to the next were to deal with possible multicollinearity effects as then assessed for all individual workers before network cohesion parameters are often correlated and after treatment. A multinomial model was to each other. We selected the two first principal then carried out with the transitions (i.e. pos- components for the subsequent analysis since sible progressions patterns from a previous to they explained 72.8% of the observed variance. 13 The effect of forager loss on honeybee temporal polyethism Page 5 of 13 51 The normality of the residuals was checked with foragers. After the removal of 50% of the forag- Shapiro–Wilk tests. Subsequently, we ran linear ers (that were older than 30 days), it is possible models (package stats version 4.0.2) with the to observe a shift in the proportion of bees per- two primary components coded as response vari- forming all tasks, except for the broodcare which ables, time of treatment (before or after manipu- was unaffected with 21% (n = 383) of the work- lation) and hive id as fixed co-factors to test for ers still performing this task. For the remaining the impact of treatment on the social network tasks, we observed a decrease in the proportion structure of the colonies. ANOVA’s were used bees performing nest maintenance (34%, n = 616) to test for overall differences of network metrics and guarding the nest (4%, n = 73), while there primary components among hives and treatments was an increase in the proportion of food storers (package car version 3.0-13). (39%, n = 725) and foragers (2%, n = 41). There is, therefore, a clear pattern of workers adapting their task allocation with respect to the different age 3. RESULTS groups as a response to the sudden loss of forag- ers in the hive (multinomial model, F4,20 = 15.572, 3.1. Effect of forager loss on task p < 0.001) (Figure 1). The specific tasks that were allocation distribution significantly affected by our treatment was guard- ing behaviour (Tukey post-hoc test, t =  − 7.376, Before the experimental manipulation, the p < 0.001, df = 20) and food storage (Tukey post- percentage distribution of worker bees aged hoc test, t = 2.417, p < 0.025, df = 20). between zero and 30 days that were performing An additional aspect of our study was to specific tasks in the hive was the following based analyse if the time taken for bees to transition on the total number of behavioural observations: from one task to the next would be affected by nest maintenance 42% (n = 515), broodcare 21% the experimental removal of foragers. Indeed, (n = 253), food storage 28% (n = 343), guarding there is a strong overall effect on the proportion the nest 9% (n = 116) and only 0.2% (n = 3) of and average time that workers took to transition Figure 1.  The proportion of task allocation groups according to age before and after the experimental manipulation (removal of foragers) based on a multinomial model fitted to the observed frequency of tasks over time in A. mellif- era. The data shows that worker bees adapt their distribution as a response to an abrupt loss of foragers. 13 51 Page 6 of 13 E. Beck et al. among the different tasks before and after the 3.2. Effect of treatment on social experimental manipulation (multinomial model, network dynamics F8,32 = 1291.79, p < 0.001). More specifically, there were significant differences on the transi- We performed nine social network analyses tion from broodcare to food storage (Tukey post- for each of the treatment hives (4 days before, the hoc test, t =  − 2.837, p = 0.008, df = 32) and from day of the treatment, and 4 days after treatment). nest maintenance to guard (Tukey post-hoc test, Overall, networks become more closely clustered t = 2.097, p = 0.044, df = 32). It is worth mention- after treatment and they also present higher k-core ing two not significant task transitions that did scores, particularly for hives B and C (Figure 3). not occur before the treatment that were from The latter hive also shows high number of nodes nest maintenance directly to foraging behaviour, in the inner and outer k-cores without many bees completely skipping several tasks under regular in the middle k-core. Overall, the hives showed an circumstances (Tukey post-hoc test, t =  − 1.753, average interaction rate of 2.55 and had a k-core p = 0.089, df = 32) and from food storage to for- index of 2.5 before the experimental manipula- ager (Tukey post-hoc test, t =  − 0.805, p = 0.427, tion, whereas after treatment the values increased df = 32) (Figure 2). In addition to completely to 3.74 and 3.6, respectively. Visual inspection of skipping some tasks, workers also performed the specific task groups in the networks before and quicker transitions. The workers shifting from after manipulation shows that different tasks are broodcare to food storage decreased in propor- well represented at both the periphery and centre tion but transitioned 2 days earlier and food stor- of the networks, indicating that workers with high ers advanced their transitioning time to guards by and low levels of interaction are equally repre- 1.6 days after treatment. Some broodcare work- sented within each type of task in the colony. The ers skipped the task of food storage and made the Wiener index followed the same trend as the aver- transition to guards also half a day quicker after age interaction rate and K-core index, increasing the removal of foragers. after treatment from 7137 to 32055. Figure 2.  Workers adjust both the proportion and time taken to transition to a new task when faced with a substan- tial loss of foragers as seen by the outcome of a multinomial model fitted to the relative frequency of task progression in function of the time before and after the treatment in A. mellifera. 13 The effect of forager loss on honeybee temporal polyethism Page 7 of 13 51 Figure 3.  Social network structures of the three experimental hives A, B and C before and after the removal of foragers. The colours of the nodes correspond to task workers were performing and tie colours represent the type of interaction. The larger the k-core, the larger is the node. Foragers are not present because the observations are pri- marily focused on tasks within the nest. 13 51 Page 8 of 13 E. Beck et al. The total variance on the social network met- groups, we observed most workers following a pat- rics explained by the two first principal com- tern of temporal division of labour (Seeley 1995; ponents was of 72.8% (PC1 38.8% and PC2 Seeley and Kolmes 1991; Wilson 1971). Both 34%, Figure 4a). Nevertheless, only the second before and after the treatment, the temporal pattern component differed significantly between the of task allocation was in agreement with previous period before and after the treatment (ANOVA, finding in the literature whereby newly emerged F 1,23 = 23.476, p < 0.001, Figure 4b), which bees engage mostly in nest maintenance tasks and was mostly determined by the average interac- broodcare, shifting to storing food, guarding the tion rate, K-core index, and Wiener index. The nest and ultimately becoming foragers (Johnson first component was mostly influenced by the 2008; Seeley 1995; Seeley and Kolmes 1991). diameter, density, average interaction rate and Guarding and foraging behaviour is mostly seen K-core index though not significantly (ANOVA, in older bees because these are the riskiest tasks. F1,23 = 1.533, p = 0.228). Hence, by having the older bees performing these tasks, colonies are able to maintain a good over- all fitness as these workers are at the end of their 4. DISCUSSION lifespan and a loss of young bees could otherwise severely impair colony functioning (Tofilski 2002). 4.1. Effect of treatment on task allocation Interestingly, the proportions of workers per- forming food storage and guarding tasks were Our results show that the task allocation in most affected by our treatment, resulting in an the honeybee A. mellifera was impacted not only increase of food storers but a decrease in guards by age but also by the proportion of individuals after treatment. There was also a trend, though performing each specific task in the hive. Despite not significant, of an increase in the abundance some bees becoming specialized in certain task of foragers after the experimental manipulation. Figure 4.  Principal component analysis of the social network metrics. a Primary component analysis shows that some network measures are more affected than others when colonies are faced with a significant loss of foragers. Dif- ferent colours represent social network analyses carried out before (cyan) and after (red) the experimental manipula- tion. b Boxplots demonstrating the significant effect observed in PC2 (p < 0.001) in function of the treatment, mostly caused by variations on the network average interaction rate, K-core index and Wiener index. 13 The effect of forager loss on honeybee temporal polyethism Page 9 of 13 51 The decrease in the proportion of guards might treatment. The principal component analysis be linked with the fact that workers skipped the showed that treatment caused a significant guarding task completely, transitioning imme- change in its second dimension which was diately to a foraging role to supply the sudden mostly determined by the increase of the demand for new foragers. There was no imme- average interaction rate, the K-core and the diate significant effect of treatment on workers Wiener indices. Both the average interaction performing nest maintenance or broodcare tasks. rate and K-core index undergo similar changes This is logical from a resource allocation per- through the treatment. This pattern can be spective (Schmickl and Crailsheim 2007), since explained by the strong positive correlation the removal of foragers does not immediately present between nodal coreness and interaction impact the brood condition of the hive i.e. the rate (Shin et al. 2018). An increased average demand of cleaning the rearing cells and feeding interaction rate corresponds to an increased the brood remains and the survival of the colony number of interactions by every actor in the partly relies on the survival of the progeny (Lee network. The removal of foragers increased the and Winston 1987). Nevertheless, the manipu- average interaction rate of the social networks lation does impact the food storing which is meaning that, on average, workers interacted directly linked with the foraging task partitioning with more nestmates when compared to before as well as guarding which precedes the foraging the manipulation. A higher connectedness inside role (Anderson and Ratnieks 1999; Ratnieks and the network leads to a faster flow of information Anderson 1999; Seeley 1995). among all actors in the network (Mersch 2016). Workers making the transition from brood- A colony may quickly adapt to changing external care to food storage decreased substantially in and internal conditions, and our results suggest their proportion through treatment, and a sig- that colonies might increase their information nificant increase was seen in the proportion flow to swiftly communicate the sudden loss of workers moving from nest maintenance to of foragers to all nestmates as fast as possible guarding. This observation supports our previ- (Biesmeijer and Slaa 2004; Franks et al. 2002; ous observation, showing that, workers will react Seeley 1986). Social insects are well-known for by skipping tasks like broodcare and food stor- their plasticity in interaction networks. Notably age and immediately transition to more external in the context of famine relief in which starved tasks such as guarding and foraging in response ant colonies showed an increased in the overall to the removal of foragers. Our data shows that number of interactions i.e. average interaction before the experimental manipulation workers rate and decrease in spatial fidelity or workers performing broodcare gradually proceed to food (Sendova-Franks et al. 2010). In addition, social storage. However, a similar proportion transition insect colonies are able to adapt the topology from nest maintenance to guarding after treat- of their interaction network to reduce disease ment, despite the greater age difference observed transmission by increasing the modularity between these tasks under regular age polyethism and clustering of the network (Stroeymeyt (Seeley 1995; Seeley and Kolmes 1991; Wilson et al. 2018), or by favouring interactions between 1971). In addition to completely skipping some specific age or task cohorts (Naug and Smith tasks, workers also performed quicker transition 2007; Naug 2008; Quevillon et al. 2015). between tasks, ranging from half a day to 2 days In addition, the manipulation caused the earlier when compared to before manipulation. social networks to increase their number of cores as a consequence of individuals that were more 4.2. Social network dynamics closely connected to others after the treatment. The K-Core index gives an indication of the Our social network analyses indicated stability of a network against possible decay that some network cohesion metrics are (Pusceddu et al. 2021). It has been shown that significantly affected by the experimental the distribution of the k-core network has an 13 51 Page 10 of 13 E. Beck et al. influence in the network resilience and that Robinson et al. 1994). Foragers are fundamen- networks with a high number of individuals tal for the collection of food resources and the in the most centralized and peripheral overall homeostasis and survival of the colony k-cores are better protected against network (Prado et al. 2020). Without a strong foraging deviations (Burleson-Lesser et al. 2020). The force, a colony would likely be weakened and removal of foragers could have prompted ultimately collapse. It is therefore paramount the colonies to adapt to this disturbance and that the workers, as showed by our results, act strengthen themselves as a response to possible quickly in restoring a well-established forag- future losses. Our data is consistent with ing role in the colony. Indeed, social network previous findings where the resilience of the parameters also changed after the experimental communication network was maintained after treatment with the removal of foragers causing experimental removal of foragers (Gernat et al. the colony to develop a more connected inter- 2018). Another explanation could be linked with action network suggesting an increased resil- the fact that the colonies were just in a phase of ience. Even when faced with a substantial per- colony expansion and that the higher K-cores are turbation such as losing half of their foragers, a result of the increased growth of the colony. hives can react and swiftly adapt to these situ- This could also explain the increase in the ations maintaining homeostasis and survival of Wiener index, which is somewhat contradictory. the colony, which might suggest that they may An increased Wiener index after treatment might be resilient to the likely uncertain climatic sce- be related to more openly clustered structures narios of the future. (Scardoni and Laudanna 2011) or, alternatively, be an artifact of the increase in the number of AUTHOR CONTRIBUTION nodes (Tian and Choi 2013). Nevertheless, when analysing specific task groups in the networks RCO and TW conceived the original idea. EB and RCO before and after removal, we can observe performed the experiments and data analyses. EB wrote the first draft of the manuscript. All authors edited and that each task is well represented in both the approved the final version of the manuscript. periphery and centre of the networks, indicating that workers with high and low degrees of FUNDING interactions are equally represented within each type of task in the colony. Indeed, the sparse Open Access Funding provided by Universitat Autonoma de sampling methodology employed in this study Barcelona. This work was supported by Research Foundation through direct behavioural observations may – Flanders (FWO) (Senior Research Grant G064120N). RCO pose quantitative limitations in capturing the is a Serra Húnter professor. multitude of social interactions within the honeybee colony. Consequently, exploring more CODE AND DATA AVAILABILITY comprehensive sampling approaches, such as The data and codes generated during the current study utilizing advanced techniques like automated are available in the Mendeley Data (Beck et al. 2022). tracking, presents an interesting avenue for future research. DECLARATIONS Overall, our results support the assumption that removal of foragers causes some workers to deviate from the normal sequence of age Ethics approval According to Belgian law, research with polyethism and transition quicker or even com- insects is exempt from ethics committee approval. pletely skip tasks in order to compensate for a Consent to participate Not applicable. sudden loss of foragers. This behaviour is an adaptation to the precarious situation in which Consent for publication Not applicable. a colony is in when it has lost a substantial part of its foragers (Huang and Robinson 1996; Competing interests The authors declare no competing interests. 13 The effect of forager loss on honeybee temporal polyethism Page 11 of 13 51 Open Access This article is licensed under a Creative Society of London. Series B: Biological Sciences Commons Attribution 4.0 International License, which 357(1427):1567–1583 permits use, sharing, adaptation, distribution and repro- Gadagkar R (1987) What are social insects? IUSSI duction in any medium or format, as long as you give Indian Chapter Newsletter 1(2):3–4 appropriate credit to the original author(s) and the source, Gernat T et al (2018) Automated monitoring of behav- provide a link to the Creative Commons licence, and ior reveals bursty interaction patterns and rapid indicate if changes were made. The images or other third spreading dynamics in honeybee social networks. party material in this article are included in the article’s Proc Natl Acad Sci 115:1433–1438 Creative Commons licence, unless indicated otherwise in Grandjean M, Jacomy M (2019) Translating networks: a credit line to the material. If material is not included in assessing correspondence between network visu- the article’s Creative Commons licence and your intended alisation and analytics. In Digital Humanities Con- use is not permitted by statutory regulation or exceeds the ference. Utrecht. Retrieved from https://​h alshs.​ permitted use, you will need to obtain permission directly archi​ves-​ouver​tes.​fr/​halshs-​02179​024 from the copyright holder. To view a copy of this licence, Grimaldi D, Engel MS (2005) Evolution of the insects. visit http://creativecommons.org/licenses/by/4.0/. Florida Entomologist, vol. 90. Cambridge Uni- versity Press. https:// ​ d oi. ​ o rg/ ​ 1 0. ​ 1 653/ ​ 0 015-​ 4040(2007)​90[588:​eoti]2.​0.​co;2 Hendrickx F, Maelfait JP, Van Wingerden W, Schweiger REFERENCES O, Speelmans M, Aviron S, Bugter R (2007) How landscape structure, land-use intensity and habi- Anderson C, Ratnieks FLW (1999) Worker allocation tat diversity affect components of total arthropod in insect societies: coordination of nectar forag- diversity in agricultural landscapes. J Appl Ecol ers and nectar receivers in honey bee (Apis mellif- 44(2):340–351. https://​d oi.​o rg/​1 0.​1 111/j.​1 365-​ era) colonies. Behav Ecol Sociobiol 46(2):73–81. 2664.​2006.​01270.x https://​doi.​org/​10.​1007/​s0026​50050​595 Huang Z-Y, Robinson GE (1996) Regulation of honey Beck E, Wenseleers T, Oliveira RC (2022) The effect of bee division of labor by colony age demography. forager loss on honeybee workers temporal poly- Behav Ecol Sociobiol 39:147–158 ethism and social network structure. Mendeley IPCC (2014) Climate Change 2014: Synthesis Report. Data, V1. https://​doi.​org/​10.​17632/​gcnst​c8syw.1 IPCC, Geneva. https://​d oi.​o rg/​1 0.​1 016/​S 0022-​ Bernadou A, Busch J, Heinze J (2015) Diversity in 0248(00)​00575-3 identity: behavioral flexibility, dominance, and Jaffé R, Dietemann V, Allsopp MH, Costa C, Crewe age polyethism in a clonal ant. Behav Ecol Socio- RM, Dall’olio, R., … Moritz, R. F. A. (2010) Esti- biol 69(8):1365–1375. https://​d oi.​o rg/​1 0.​1 007/​ mating the density of honeybee colonies across s00265-​015-​1950-9 their natural range to fill the gap in pollinator Biesmeijer JC, Slaa EJ (2004) Information flow and decline censuses. Conserv Biol 24(2):583–593. organization of stingless bee foraging. Apidologie https://​doi.​org/​10.​1111/j.​1523-​1739.​2009.​01331.x 35(2):143–157 Johnson BR (2008) Within-nest temporal polyethism in Borgatti SP, Everett MG, Freeman LC (2002) Ucinet for the honey bee. Behav Ecol Sociobiol 62(5):777– Windows: software for social network analysis. Har- 784. https://​doi.​org/​10.​1007/​s00265-​007-​0503-2 vard: MA: Analytic Technologies. Retrieved from Kevan PG, Greco CF, Belaoussoff S (1997) Log-normal- https://​sites.​google.​com/​site/​ucine​tsoft​ware/​home ity of biodiversity and abundance in diagnosis and Burleson-Lesser K, Morone F, Tomassone MS, Makse measuring of ecosystemic health: pesticide stress HA (2020) K-core robustness in ecological and on pollinators on blueberry heaths. The Journal of financial networks. Sci Rep 10(1):3357. https://​ Applied Ecology 34:1122–1136. https://​doi.​org/​10.​ doi.​org/​10.​1038/​s41598-​020-​59959-4 2307/​24052​26 Cox-Foster DL, Conlan S, Holmes EC, Palacios Lecocq A, Jensen AB, Kryger P, Nieh JC (2016) Parasite G, Evans JD, Moran NA, Lipkin WI (2007) A infection accelerates age polyethism in young honey metagenomic survey of microbes in honey bee bees. Sci Rep 6:1–11. https://​doi.​org/​10.​1038/​srep2​2042 colony collapse disorder. Science 318:283–287. Lee PC, Winston ML (1987) Effects of reproductive https://​doi.​org/​10.​1126/​scien​ce.​11464​98 timing and colony size on the survival, offspring De Joy PE (1998) The effects of low temperature on colony size and drone production in the honey metabolism and survival of individual honey bees bee (Apis mellifera). Ecological Entomology (Apis mellifera). State University of New York 12(2):187–195. https://​d oi.​o rg/​1 0.​1 111/j.​1 365-​ College at Brockport 2311.​1987.​tb009​97.x Franks NR, Pratt SC, Mallon EB, Britton NF, Sumpter Mersch DP (2016) The social mirror for division of DJ (2002) Information flow, opinion polling and labor: what network topology and dynamics can collective intelligence in house–hunting social teach us about organization of work in insect soci- insects. Philosophical Transactions of the Royal eties. Behav Ecol Sociobiol 70(7):1087–1099. https://​doi.​org/​10.​1007/​s00265-​016-​2104-4 13 51 Page 12 of 13 E. Beck et al. Michener CD (1969) Comparative social behavior of R Core Team (2020) R: a language and environment for bees. Annu Rev Entomol 14(1):299–342. https://​ statistical computing. R Foundation for Statistical doi.​org/​10.​1146/​annur​ev.​en.​14.​010169.​001503 Computing, Vienna Michener CD (2007) The bees of the world. Nuevos Ratnieks FLW, Anderson C (1999) Task partitioning sistemas de comunicación e información, 2nd edn. in insect societies. Insectes Soc 46(2):95–108. The Johns Hopkins University Press, Baltimore https://​doi.​org/​10.​1007/​s0004​00050​119 Natsopoulou ME, McMahon DP, Paxton RJ (2016) Robinson GE (1992) Regulation of division of labor in Parasites modulate within-colony activity and insect societies. Annu Rev Entomol 37:637–665 accelerate the temporal polyethism schedule of a Robinson GE, Page RE, Huang Z-Y (1994) Temporal social insect, the honey bee. Behav Ecol Socio- polytehism in social insects is a developmental biol 70(7):1019–1031. https://​d oi.​o rg/​1 0.​1 007/​ process. Anim Behav 48:467–469 s00265-​015-​2019-5 Scardoni G, Laudanna C (2011) Network centralities for Naug D (2008) Structure of the social network and its Cytoscape. Retrieved from https://​www.​centi​server. influence on transmission dynamics in a honeybee ​org/​centr​ality/​Wiener_​Index/ colony. Behav Ecol Sociobiol 62:1719–1725 Schmickl T, Blaschon B, Gurmann B, Crailsheim K Naug D, Gadagkar R (1998) Division of labor among a (2003) Collective and individual nursing invest- cohort of young individuals in a primitively euso- ment in the queen and in young and old honeybee cial wasp. Insectes Soc 45(3):247–254. https://​doi.​ larvae during foraging and non-foraging periods. org/​10.​1007/​s0004​00050​085 Insectes Soc 50(2):174–184. https://​d oi.​o rg/​1 0.​ Naug D, Smith B (2007) Experimentally induced 1007/​s00040-​003-​0644-x change in infectious period affects transmission Schmickl T, Crailsheim K (2002) How honeybees (Apis dynamics in a social group. Proc r Soc B Biol Sci mellifera L.) change their broodcare behaviour in 274:61–65 response to non-foraging conditions and poor pol- Neov B, Georgieva A, Shumkova R, Radoslavov G, len conditions. Behavioral Ecology and Sociobi- Hristov P (2019) Biotic and abiotic factors associ- ology 51(5):415–425. https://​d oi.​o rg/​1 0.​1 007/​ ated with colonies mortalities of managed honey s00265-​002-​0457-3 bee (Apis mellifera). Diversity 11(237):1–16. Schmickl T, Crailsheim K (2004) Inner nest homeostasis https://​doi.​org/​10.​3390/​d1112​0237 in a changing environment with special emphasis Panziera D, Requier F, Chantawannakul P, Pirk CWW, on honey bee brood nursing and pollen supply. Blacquière T (2022) The diversity decline in wild Apidologie 35:249–263. https://​doi.​org/​10.​1051/​ and managed honey bee populations urges for an apido:​20040​19 integrated conservation approach. Front Ecol Evol Schmickl T, Crailsheim K (2007) HoPoMo: a model of 10(March):1–7. https://​doi.​org/​10.​3389/​fevo.​2022.​ honeybee intracolonial population dynamics and 767950 resource management. Ecol Model 204(1–2):219– Peso M, Even N, Søvik E, Naeger NL, Robinson GE, 245. https://​doi.​org/​10.​1016/j.​ecolm​odel.​2007.​01.​ Barron AB (2016) Physiology of reproductive 001 worker honey bees (Apis mellifera): insights for Seeley TD (1986) Social foraging by honeybees: how colo- the development of the worker caste. J Comp nies allocate foragers among patches of flowers. Behav Physiol A Neuroethol Sens Neural Behav Phys- Ecol Sociobiol 19:343–354 iol 202(2):147–158. https:// ​ d oi. ​ o rg/ ​ 1 0. ​ 1 007/​ s00359-​015-​1061-0 Seeley TD (1995) The wisdom of the hive: the social physiology of honey bee colonies. Harvard Uni- Potts SG, Biesmeijer JC, Kremen C, Neumann P, Schweiger versity Press O, Kunin WE (2010) Global pollinator declines: trends, impacts and drivers. Trends Ecol Evol Seeley TD, Kolmes A (1991) Age polyethism for hive 25(6):345–353. https://​doi.​org/​10.​1016/j.​tree.​2010.​ duties in honey bees—illusion or reality? Ethology 01.​007 87:284–297 Prado A, Requier F, Crauser D, Le Conte Y, Bretagnolle Sendova-Franks AB et al (2010) Emergency network- V, Alaux C (2020) Honeybee lifespan: the criti- ing: famine relief in ant colonies. Anim Behav cal role of pre-foraging stage: Honey bee lifespan. 79:473–485 Royal Society Open Science 7(11):1–13. https://​ Shin K, Eliassi-Rad T, Faloutsos C (2018) Patterns and doi.​org/​10.​1098/​rsos.​20099​8rsos​200998 anomalies in k-cores of real-world graphs with appli- Pusceddu M, Cini A, Alberti S, Salaris E, Theodorou P, cations. Knowl Inf Syst 54(3):677–710. https://​doi.​org/​ Floris I, Satta A (2021) Honey bees increase social 10.​1007/​s10115-​017-​1077-6 distancing when facing the ectoparasite Varroa Stout JC, Morales CL (2009) Ecological impacts destructor. Sci Adv 7(44):1–15. https://​d oi.​o rg/​ of invasive alien species on bees. Apidologie 10.​1126/​sciadv.​abj13​98 40(3):388–409. https://​d oi.​o rg/​1 0.​1 051/​a pido/​ Quevillon LE, Hanks EM, Bansal S, Hughes DP (2015) 20090​23 Social, spatial, and temporal organization in a complex Stroeymeyt N et al (2018) Social network plasticity insect society. Sci Rep 5:13393 decreases disease transmission in a eusocial insect. Science 362:941–945 13 The effect of forager loss on honeybee temporal polyethism Page 13 of 13 51 Székely T, Moore AJ, Komdeur J (2010) Social Wilson EO (1971) The insect societies. Harbard Uni- behaviour: genes, ecology and evolution. Cam- versity Press, Cambridge bridge University Press. https://​d oi.​o rg/​1 0.​1 017/​ Winfree R, Aguilar R, Vazquez DP, LeBuhn G, CBO97​8 0511​7 81360 Aizen MA (2009) A meta-analysis of bees’ Tabassum S, Pereira F, Fernandes S, Gama J (2018) responses to anthropogenic disturbance. Ecology Social network analysis: an overview. Wiley Inter- 90(8):2068–2076 disciplinary Reviews: Data Mining and Knowledge Woyciechowski M, Moroń D (2009) Life expectancy Discovery 8(5):1–30. https://​d oi.​o rg/​1 0.​1 002/​ and onset of foraging in the honeybee (Apis mellif- widm.​1256 era). Insectes Soc 56(2):193–201. https://​doi.​org/​ Tian D, Choi KP (2013) Sharp bounds and normaliza- 10.​1007/​s00040-​009-​0012-6 tion of Wiener-type indices. PLoS ONE 8(11):1–9. https://​doi.​org/​10.​1371/​journ​al.​pone.​00784​48 Publisher’s Note Springer Nature remains neutral Tofilski A (2002) Influence of age polyethism on lon- with regard to jurisdictional claims in published maps gevity of workers in social insects. Behav Ecol Sociobiol 51(3):234–237. https://​doi.​org/​10.​1007/​ and institutional affiliations. s00265-​001-​0429-z Tofilski A (2009) Shorter-lived workers start foraging ear- lier. Insectes Soc 56(4):359–366. https://​doi.​org/​10.​ 1007/​s00040-​009-​0031-3 13