Age-related mushroom body expansion in male sweat bees and bumble bees

A well-documented phenomenon among social insects is that brain changes occur prior to or at the onset of certain experiences, potentially serving to prime the brain for specific tasks. This insight comes almost exclusively from studies considering developmental maturation in females. As a result, it is unclear whether age-related brain plasticity is consistent across sexes, and to what extent developmental patterns differ. Using confocal microscopy and volumetric analyses, we investigated age-related brain changes coinciding with sexual maturation in the males of the facultatively eusocial sweat bee, Megalopta genalis, and the obligately eusocial bumble bee, Bombus impatiens. We compared volumetric measurements between newly eclosed and reproductively mature males kept isolated in the lab. We found expansion of the mushroom bodies—brain regions associated with learning and memory—with maturation, which were consistent across both species. This age-related plasticity may, therefore, play a functionally-relevant role in preparing male bees for mating, and suggests that developmentally-driven neural restructuring can occur in males, even in species where it is absent in females.


Discussion
We found strikingly similar patterns of age-related neuroplasticity in M. genalis and B. impatiens males. In both species, expansion of mushroom body structures, including calyx and neuropil, occurred with age under experimental conditions void of ecologically-relevant experience. Our work adds to the sparse literature on volumetric neuroplasticity in Hymenopteran males while improving our general understanding of potential functions of experience-expectant neuroplasticity in insects.
Our results provide the first definitive evidence that large-scale volumetric neuroplasticity is driven by age, independent of relevant experience (e.g., social, flight, etc.), in male Hymenoptera. Previous research has documented mushroom body expansion associated with the combination of maturation and experience in paper wasp 14 and honey bee 13 males. Similar to our results, paper wasp and honey bee calyx and neuropil volumes increase, respectively, as males mature. Developmental maturation was observed with flight initiation, social interaction, and mating behavior in these species 13,14 . But, since the individual effects of age and experience were not experimentally controlled for in these previous studies, they could not be evaluated independently. Mature males in our study were experimentally deprived of flight, social cues, and mating experience. We aimed to isolate experience-independent from experience-dependent brain changes during adult development. However, social deprivation can adversely affect eusocial insects, leading to impaired brain development, learning, and behaviors [30][31][32] . Therefore, we cannot eliminate the possibility that the volumetric plasticity observed may include effects associated with unnatural rearing conditions, inadvertent stress, or unidentified experiences. Nevertheless, our results show that, across multiple species with different rearing conditions, male brains change relatively consistently with age, independent of ecologically-relevant experience, suggesting that age-related neuroplasticity may be common in male Hymenoptera.
Age-related mushroom body expansion coincides with reproductive maturation in male bees, and may represent a common developmental change associated with dispersal from the nest prior to the onset of mating. A primary function of male Hymenoptera is to inseminate a female(s) 17,33-37 , and experience-expectant neuroplasticity may potentially facilitate this behavior. Our approach cannot identify precisely when age-related brain changes occurred in either species. However, bumble bee (B. vosnesenskii) males reach reproductive maturity by 8-10 days post-eclosion 38 , and we observed significant mushroom body expansion in B. impatiens males after 10 d of aging. Scent-marking and patrolling is the most common pre-mating strategy in Bombus 27 , whereby males pheromonally mark points along a flight route 27,39,40 . Patrolling is similar to "trap-line" foraging 27 , which is associated with the phylogenetic expansion of mushroom bodies in Heliconius butterfly species that also exhibit age-related brain plasticity 41 . Female bumble bees utilize learned aspects of their environment for spatial orientation 23,24 , and males have learning capabilities equivalent to females 42 ; therefore, while speculative, the brain changes observed with Bombus male maturation may be an important preparation for the potential cognitive challenges related to mate finding behaviors. Similarly, we found that mature (6-d old) sweat bee males had enlarged mushroom bodies relative to newly eclosed males. The mating behavior of M. genalis is unknown. However, males typically stay in their natal nest for up to 4 d past emergence 28 , during which time they are fed via trophallaxis by their mothers and sisters 43 . It is presumably during this time that they are becoming reproductively mature. The males of some Halictine species exhibit mate patrolling 44 , but it is unknown whether M. genalis conduct these behaviors. In honey bees, neuropil expansion coincides with the time that males reach sexual maturity (6-12 d) 13,45,46 . Age-related neuroplasticity observed in male paper wasps (M. mastigophorus) may also coincide with reproductive maturity 14 . Males of this species are atypical of other social insects in that www.nature.com/scientificreports/ they remain on their natal nests long after eclosion, departing only temporarily to mate 47,48 . The age at first nest departure (median = 5 d 49 ), however, is still comparable with those of the bees studied here. Our study was not designed to identify the functional relevance of age-related brain development in males, but instead provides new insight for subsequent work. Thus, while the functional roles of age-related plasticity remain unclear, our study and previous studies suggest that the age-related neuroplasticity observed within males across species may be associated with departing the nest in search of mating opportunities-a predictably timed, common event driving the male life-cycle. This pattern of expansion is similar to the 'experience-expectant' neuroplasticity observed in the females of some, but not all, social insects.
Our results also suggest that intraspecific sex differences in age-related neuroplasticity patterns can occur among some social insect species. Neuroanatomical changes in the female workers of highly social bees accompany shifts in colony needs [6][7][8][9][10][11] , coinciding with age-related behavioral transitions from working inside the hive to foraging 6,9,50 . Yet, this age-related task specialization is not universal across social species. In bumble bees, where division of labor is size-based instead of age-related, females exhibit mushroom body expansion within the first few days of life 12,51,52 , which accompanies their capacity for behavioral maturation soon after emergence 53 . These changes are similar to those observed in our mature B. impatiens males. However, experience-expectant neuroplasticity is absent in M. genalis females 54 (though see 55 ), which also lack age-related task specialization 25,56 . Our finding that mushroom body expansion occurs with maturation in male M. genalis suggests that experience-expectant neuroplasticity can occur in males, even when it is absent in females. Future work comparing sex-specific patterns of brain development in additional bee species is needed to determine the pervasiveness of intersexual differences in neuroplasticity. Investigating socioecological drivers of neuroplasticity in both sexes, particularly in solitary species where females lack age-related plasticity 57,58 , will provide a more robust understanding of the relationship between neuroplasticity and social evolution.

Field collections and laboratory rearing. We conducted the experiment for Megalopta genalis from
March to May 2015 on Barro Colorado Island (BCI), Republic of Panama. Twice daily-once in the morning and evening-we collected newly eclosed males from their natal nests, which consisted of dead sticks or branches 25 . We randomly assigned newly eclosed males to either 'newly eclosed' or 'mature' treatment groups. Bees designated as 'newly eclosed' (N = 8) were sacrificed within minutes, whereas 'mature' (N = 6) males were housed individually in food storage containers for 6 d in an incubator (27 °C, 70%, 0:24 L:D) and provided food (36% sugar, 7% protein w/v) ad libitum. Food was mixed by dissolving six Nature's Blend Protein tablets (National Vitamin Company, Casa Grande, AZ) in 50 ml distilled water and changed twice daily.
During August to December 2018, we produced B. impatiens males from queenless microcolonies (N = 11) generated using three commercial colonies from Koppert Biological Systems (Howell, MI, USA). Microcolonies consisted of five B. impatiens workers from the same source colony that were housed in custom rearing cages: 173 × 130 × 91 mm food storage containers that included aluminum mesh bottoms and hinged plexiglass tops. We supplied microcolonies with 50% sugar water (cane sugar dissolved in distilled water) supplemented with potassium sorbate, citric acid, Honey B Healthy Essential Oil, and Honey B Healthy Amino Boost, as well as pollen dough ad libitum. Our pollen dough consisted of honey bee collected pollen (Betterbee, Greenwich, NY, USA) mixed with the aforementioned sugar water until it reached a consistency similar to moist, slightly tacky fine-grained sand. We stored microcolonies in an incubator maintained at 27 °C and ~ 60-70% relative humidity on a 16:8 h light:dark cycle. Male brood require ~ 24 d to develop before eclosion 59 . We checked brood development and for newly eclosed males daily. After eclosion, we randomly assigned new males to one of two treatment groups: 'newly eclosed' (N = 11) males were sacrificed immediately, whereas 'mature' (N = 7) males were maintained individually for 10 d in the rearing cages and conditions described above.
Where applicable, we followed the recommended guidelines for animal care and use 60 .
Confocal microscopy and structure tracing. We imaged brains using autofluorescence on a laser confocal microscope (Zeiss LMS 710). Whole brains were mounted in methyl salicylate and scanned as z-stack series ranging from 760 to 925 µm thick. We imaged brains as 3 × 2 tile scans (2867 × 1946 pixels) with optical slices captured in 5 µm intervals. For both species, brains were imaged simultaneously using two lasers, though the wavelengths, laser power, and gains varied by species. We imaged M. genalis at 410-484 nm and 495-538 nm wavelengths, 3.5 and 3.0 power, and 504-535 and 495-517 gains for laser 1 and 2, respectively. For B. impatiens, the first laser had a wavelength between 410 and 485 nm, a laser power of 4.0, and a gain range between 527 and 567. The second laser had a wavelength, power, and gain range of 495-538 nm, 3.5, and 518-558. Whole brain image stacks were saved as individual jpegs. www.nature.com/scientificreports/ Throughout confocal stacks, we traced individual structures on every other optical slice (10 µm intervals) and estimated volumetric measurements via serial reconstruction using Reconstruct software (Fiala 62 ; version 1.1.0.0; available at http:// synap ses. clm. utexas. edu). Due to occasional tissue damage, we traced each structure unilaterally to maximize sample inclusion. For undamaged brains, we randomly selected either the right or left side to trace, whereas undamaged sides were always traced for brains with tissue damage. The number of right and left side brain traces were distributed similarly across treatment group (M. genalis: Yates corrected X 2 (1, N = 14) = 0.29, p = 0.589 and B. impatiens: Yates corrected X 2 (1, N = 16) = 0.02, p = 0.896). Whole brain traces were also conducted unilaterally, corresponding with the side used for structure tracing, and always excluded the lamina and retina (Fig. 3). We conducted all confocal imaging, tracing, and 3D reconstruction without knowledge of the experimental treatment group to which each sample belonged.
The structures examined included the calyces (lip, collar, and basal ring as one structure), mushroom body lobes (peduncle and lobes as one structure), total neuropil (calyces and mushroom body lobes), and Kenyon cells (Fig. 3). Neuropil to Kenyon cell volumetric increases can occur with age-related plasticity 6,7,29,57 ; therefore, we also assessed neuropil:Kenyon cell ratios (N:K). For each sample, we standardized volumes using two methods (see Supplementary Information online): 1) structure volumes to whole brain, referred to as "relative volumes", and 2) structure volumes to Kenyon cells (results reported in Supplementary Information online).
We assessed the relative volumes (structure:wholebrain), structure:KC , and N:K ratios using Student's t-tests (stats, version 4.0.4) to compare 'newly eclosed' and 'mature' bees independently for each species. We used visual inspection of qq-plots (car, version 3.0-10; Fox and Weisberg 63 ) and Anderson-Darling normality tests (Nortest, version 1.0-4; https:// CRAN.R-proje ct. org/ packa ge= norte st) to verify normality assumptions. One variable-relative calyx volume for B. impatiens-violated normality assumptions, so we applied a Box-Cox transformation (MASS, version 7.3-53; Venables and Ripley 64 ) using λ = − 1.455. We assessed homogeneity of variance using R package car (version 3.0-10; Fox and Weisberg 63 ). Relative Kenyon cell volume violated variance assumptions, therefore we conducted a second Box-Cox transformation using λ = − 0.970. We determine effect size between groups by calculating Hedges' g (effsize, version 0.8.1, Torchiano 65 ). To account for multiple comparisons, we applied a Bonferroni correction and adjusted statistical significance to α = 0.01.