Cuckoo male bumblebees perform slower and longer flower visits than free-living male and worker bumblebees

Cuckoo bumblebees are a monophyletic group within the genus Bombus and social parasites of free-living bumblebees, upon which they rely to rear their offspring. Cuckoo bumblebees lack the worker caste and visit flowers primarily for their own sustenance and do not collect pollen. Although different flower-visiting behaviours can be expected between cuckoo and free-living bumblebees due to different biological constraints, no study has yet quantified such differences. Here, we provide the first empirical evidence of different flower-visiting behaviours between cuckoo and free-living bumblebees. We recorded the flower-visiting behaviour of 350 individual bumblebees over two years in a wild population of the entomophilous plant Gentiana lutea, of which they are among the main pollinators. In cuckoo bumblebees (28.9% of the total), we only found males, while we found both workers and males in free-living bumblebees. Cuckoo bumblebees visited significantly more flowers for longer time periods than both free-living bumblebee workers and males within whorls, while differences at the whole-plant level were less marked. Free-living bumblebee males visited more flowers and performed slightly longer flower visits than workers. Behavioural differences between cuckoo male bumblebees and free-living bumblebee workers are likely related to different foraging needs, while differences between cuckoo and free-living bumblebee males may be caused by differences in colony development and a delayed mating period of free-living bumblebees. The longer visits made by cuckoo male bumblebees will likely negatively affect plant reproductive success through increased within-plant pollen flow.


Introduction
Bumblebees are primitively eusocial bees with an annual life cycle usually divided into three distinct phases.
First, a solitary phase involving hibernation of mated gynes followed by nest foundation (from late summer to early spring), then a multiplicative phase during which the founding queen produces workers (from mid-spring to mid-summer), and finally a reproductive phase during which males and gynes are produced (mid-late summer) (Alford 1975;Lhomme and Hines 2019).
As in other groups of social insects, social parasitism has evolved in several species of bumblebees, called cuckoo bumblebees (Lhomme and Hines 2019). There are currently 27 species of cuckoo bumblebees (subgenus Psithyrus) recognized worldwide (Lhomme et al. 2021). These species completely lack the worker caste (i.e., all the female individuals are fertile females) and do not have specialized structures on the hind legs (i.e., corbiculae) to collect pollen to feed their larvae. Consequently, females of cuckoo bumblebees usurp the nests of host social bumblebees (hereafter referred to as free-living bumblebees) almost always killing the host queen to force the host workers to rear their offspring (Fisher 1987).
Bumblebees are important pollinators of both wild flowering plants and crops (e.g., Thomson and Goodell 2001;Pellissier et al. 2012;Fründ et al. 2013;Andrikopoulos and Cane 2018), and a few species are exploited commercially in agriculture (Velthuis and van Doorn 2006;Goulson et al. 2008). As most pollinators, free-living bumblebees show intraspecific differences in foraging behaviour between sexes mainly in relation to offspring provisioning (Smith et al. 2019). Workers visit flowers to collect pollen to feed the larvae and nectar for energy intake, while males visit flowers to seek nectar to fuel their flight and to look for mates (Goulson 2010).
Consequently, foraging females have generally high levels of flower constancy (Russell et al. 2017), which increase pollination efficiency by transferring conspecific pollen between flowers of a same species (Goulson 2010). However, male bumblebees can also inadvertently carry large amounts of pollen that can potentially contribute to pollination (Jennersten et al. 1991;Ostevik et al. 2010;Wolf and Moritz 2014). Although flower visiting behaviour of free-living bumblebees has been widely studied, behaviour of cuckoo bumblebees remains almost unstudied (but see Bateman and Rudall 2014), and to our knowledge there is still no assessment of similarities or differences between parasitic and non-parasitic bumblebees.
In this article, we investigate the flower-visiting behaviour of cuckoo and free-living bumblebees in a wild population of the entomophilous Gentiana lutea subsp. symphyandra, where both bumblebee types were abundant and among its most important pollinators. Specifically, we aim to answer two main questions: (1) do cuckoo and free-living bumblebees differ in the time spent visiting flowers and in the number of flowers visited, and (2) do cuckoo bumblebees display different foraging behaviours from those of males and workers of freeliving bumblebees? An evaluation of the different flower-visiting behaviours between parasitic bumblebees and their hosts would increase our knowledge of their biology and allow us to estimate differences in pollination efficiency with implications for plant fitness.

Study site and species
This study was carried out in a natural population of Gentiana lutea L. subsp. symphyandra (Murb.) Hayek situated on the eastern face of Mount Grande (Northern Apennines; Bologna, Italy), within the Habitat Directive site "IT4050002 SIC-ZPS Corno alle Scale" (1380-1460 m a.s.l.; 44° 8′ 57″ N, 10° 52′ 10″ E). The study population is located in a steep clearing surrounded by a beech (Fagus sylvatica L.) forest.
Gentiana lutea subsp. symphyandra is a perennial herb growing on mountains of south-eastern Europe that flowers between June and July (Rossi et al. 2015). The tall fertile stems bear several yellow flowers with wideopen corollas of about 4 cm in diameter (personal observation MG), grouped in pseudo-whorls (hereafter: whorls) that flower sequentially from the bottom to the top. Flowers of G. lutea are functionally and ecologically generalized (sensu Ollerton et al. 2007), being efficiently visited by several insect species belonging to four orders (Hymenoptera, Diptera, Coleoptera, Lepidoptera) with different physiological and energetic requirements (Rossi et al. 2014). Among these, bumblebees have the highest pollinator performance, fidelity and visitation frequency in the studied population (Rossi et al. 2014). Although G. lutea is a self-compatible species, the partial flower dichogamy reduces within-flower selfing, while spontaneous or pollinator-mediated selfing within plant (i.e. geitonogamy) is more likely to occur (Rossi et al. 2014). Seeds developed from selfpollination have lower viability and germination than cross-pollinated seeds (Rossi et al. 2016).

Bumblebee observations and sampling
Field observations were carried out over a total of six days in two non-consecutive years (three days in both 2013 and 2015), between July 14 th and July 23 rd each year, during the flowering peak of G. lutea subsp.

Data analysis
Since the focus of this article is not to describe temporal patterns of changes and because we did not find significant differences between years in a preliminary analysis, we pooled data from 2013 and 2015 in all analyses. Moreover, in a preliminary analysis we did not find significant differences between species in either the time spent on flowers or the number of flowers visited, therefore we pooled species data within cuckoo and free-living bumblebee types. We used Pearson's product-moment correlation coefficients to test for correlations between the number of flowers visited and the time spent visiting flowers within whorls and plants by bumblebees.
To test whether bumblebees showed differences in the time spent visiting flowers we used linear mixed models (LMMs). We used the log-transformed time (measured in seconds) spent foraging on flowers by individual bumblebees as response variable, and bumblebee type (i.e., males of cuckoo and free-living bumblebees, workers of free-living bumblebees) as fixed effect. We tested two separate LMMs for the time spent within whorls and the time spent on the whole plant. In addition, we fitted two different LMMs including the number of whorls displayed by plants as additive factor or in interaction with bumblebee type to test for effects on the time spent on flowers at the whole-plant level. We included the observation interval, nested within day and year, as a random effect to account for temporal changes. In LMMs on the time spent on the whole plant, the best models in terms of goodness-of-fit were chosen based on the Akaike Information Criterion with correction for small sample sizes (AICc), selecting the model(s) with the lowest AICc value (Burnham and Anderson 2002). If ΔAICc values for some next best models were lower than 2, we used model averaging to calculate a weighted average of the parameter estimates (Burnham and Anderson 2002). After selecting the best models we performed pairwise comparisons between bumblebee types by estimating leastsquares means with Tukey-adjusted p-values (Lenth 2020).
To evaluate differences in the number of flowers visited by bumblebees we fitted generalized linear mixed models (GLMMs) with Poisson error distribution and with the numbers of visited flowers as response variable, bumblebee type as fixed effect and observation intervals nested within day and year as random effects. We tested two separate GLMMs for the number of flowers visited within whorls and the number of flowers visited on the whole plant, and then performed pairwise comparisons between bumblebee types.

Results
We observed a total of 350 bumblebees during 7.5 hours of observations over the two years of sampling, corresponding to three species of cuckoo bumblebees and four species of free-living bumblebees (Table 1).
Bombus rupestris was the most abundant species among cuckoo bumblebees, while B. terrestris and B.
lapidarius were the most abundant species among free-living bumblebees (Table 1). Overall, we only found cuckoo bumblebee males, and workers were more frequent than males in free-living bumblebees (ratio 1.6:1).
Cuckoo bumblebees visited significantly more flowers per whorl (4.74 ± 0.4) than free-living bumblebee workers (2.67 ± 0.2), but no more than males (4.73 ± 0.6), while free-living bumblebee males visited significantly more flowers than workers ( Figure S1A, Table S4). We did not find significant differences in the number of flowers visited at the whole plant level between bumblebee types ( Figure S1B, Table S5).

Discussion
In this article we explored differences in the flower-visiting behaviour of cuckoo and free-living bumblebees.
We provide the first empirical evidence of slower and significantly longer flower visits performed by cuckoo bumblebees compared to free-living bumblebees. We found that male cuckoo bumblebees visited more flowers and spent more time during each visit than both free-living bumblebee workers and males at the whorl level, but these differences were less marked at the plant level. Free-living bumblebee males visited more flowers than workers but not for significantly longer time periods.

Flower-visiting behaviour
The different behaviours observed between cuckoo bumblebee males and free-living bumblebee workers can be mainly related to different foraging needs. Bumblebee workers need to collect pollen for colony development, while at the same time keeping a positive balance between energy intake (through nectar) and the energy spent visiting flowers (Hodges and Wolf 1981;Pyke 1984). By contrast, cuckoo bumblebee males, as free-living males, do not collect pollen and visit flowers mainly for nectar. Consequently, the need to optimise energy intake over expenditure during foraging bouts is probably less important for cuckoo bumblebees than for free-living bumblebee workers.
Bumblebee workers visited fewer flowers for slightly shorter periods than free-living males. Though differences in visit duration were not statistically significant, mainly because of the high response variability of free-living bumblebee males, the different observed behaviours remain biologically relevant and support previous literature (Ostevik et al. 2010;Wolf and Chittka 2016). Shorter visits made by workers are likely driven by different foraging needs compared to males, as males do not collect pollen to sustain the colony similarly to cuckoo male bumblebees. Moreover, visits made by workers are usually driven by the quality and abundance of floral rewards, as workers tend to prioritize highly rewarding flowers to collect both pollen and nectar to optimize foraging bouts, while male visits are less constrained by reward availability (Smith et al. 2019).
The longer visits and the higher number of flowers visited within whorls by cuckoo bumblebees cannot be attributed to different foraging requirements compared to free-living bumblebee males since neither of them provide pollen for the nest and mainly visit flowers for their own sustenance and to look for mates (Goulson Corbet 1991), we frequently observed almost lethargic movements by cuckoo bumblebees during flower visits within whorls of G. lutea. We hypothesize that differences between cuckoo and free-living male bumblebees can be linked to different mating periods. Cuckoo bumblebee females tend to usurp free-living bumblebee nests at the beginning of colony development, when only the few workers of the first brood have been produced (Kreuter et al. 2012;Lhomme et al. 2013). Moreover, cuckoo bumblebees start to lay male and female eggs within 10 days from invasion (Lhomme et al. 2013), while free-living bumblebees start to lay male eggs between 30 and 40 days after the emergence of the first brood, and the last group of diploid eggs laid before male eggs usually develops into gynes (Bogo et al. 2018). However, the developmental time of cuckoo and free-living bumblebee sexual morphs (males and gynes) is comparable (Küpper and Schwammberger 1995;Bogo et al. 2018). Consequently, a gap between the mating periods of cuckoo and free-living bumblebees can be expected. In our observations we only found cuckoo bumblebee males, suggesting that the peak flowering of G. lutea occurred when the mating period of cuckoo bumblebees was toward the end, therefore reducing the need for males to actively look for partners. By contrast, free-living bumblebees were likely still at the beginning of their mating period, compelling free-living male bumblebees to actively seek out partners and increase their mobility. In addition, cuckoo and free-living bumblebees may greatly differ in mating duration (e.g., 3 min versus more than 26 min, respectively; Lhomme et al. 2013), which can potentially affect male foraging behaviour, for example through distinct energetic requirements. Differences in the metabolism of cuckoo and free-living males may further emphasize behavioural differences. Further studies are needed to explore these possibilities.

Implications for pollination
Shorter visits made by worker bumblebees tend to favour constant foraging patterns and short-term floral specialization, which in turn can translate into a more efficient pollen dispersal between plants of a same species within a given population (Smith et al. 2019 lutea.
Although we could not directly link bumblebee visits to plant fitness, we expect the longer flower visits of cuckoo bumblebees to reduce fruit set and seed set and germination. A previous study on the same population showed that self-pollinated flowers of G. lutea produced seeds with lower weight and germination rates compared to both open-pollinated and cross hand-pollinated flowers (Rossi et al. 2016). The reduced fitness can be partially explained by inbreeding depression or by intrinsic genetic problems (e.g. founder effect or bottleneck) as a consequence of the isolation from other natural populations (Rossi et al. 2016). However, the high abundance of cuckoo bumblebees observed in this study can act in addition to these factors and further reduce plant fitness in the study population. The transfer of pollen within plants facilitated by the longer visits of cuckoo bumblebees can increase geitonogamy and pollen discounting (Brunet 2005), ultimately reducing plant fitness.
Among the 350 bumblebees observed in this study, almost a third (28.9%) were cuckoo bumblebee males.
Because they are among the main pollinators of G. lutea (Rossi et al. 2014), we expect them to play an important role in the pollination success and ultimately on population persistence. Further analyses aimed directly at evaluating the effects of bumblebee visits on plant fitness would help understand the balance between positive (e.g., pollen flow) and negative (e.g., geitonogamy) contributions of cuckoos compared to free-living bumblebees. Given the widespread presence of cuckoo bumblebees, we expect that their visits can significantly affect the reproductive success of several plant species, and encourage future research to directly address this overlooked topic.

Data accessibility
Data and R scripts for analyses are available online: https://doi.org/10.5281/zenodo.5161010