Morphological characterization and staging of bumble bee pupae

Bumble bees

In this study, we examined and compared pupal as well as early adult development in both commercial and wild reared colonies of the eastern Nearctic bumble bee B. impatiens Cresson and in two other wild-reared Pyrobombus species from the western Nearctic, B. melanopygus Nylander and B. vosnesenskii Radoszkowski. Commercial colonies of B. impatiens were obtained from Koppert Biological Systems (Howell, MI, USA) and maintained prior to study under laboratory conditions (25 °C, 50% RH, total darkness). Wild bumble bee colonies were reared from queens collected from State College, Pennsylvania for B. impatiens and from Southwest Oregon and Northern California for B. melanopygus and B. vosnesenskii. These colonies were maintained in plastic boxes in the incubator (29 °C, 60% RH, total darkness). All colonies were fed with sugar water containing a preservative (Koppert Biological Systems, Howell, MI, USA) and the same source of fresh honey bee-collected pollen (Swarmbustin’ Honey, West Grove, PA, USA).

Pupa collection

Newly formed cocoon clusters were removed from colonies and a small opening was cut in the cocoons to inspect the stage of brood. Only cocoons containing prepupae or spinning last instar larvae were retained for pupal staging and duration analysis. These were placed on a plastic weigh boat in the incubator (32 °C, 60% RH, total darkness) and were checked every 4 h for pupal ecdysis, as prepupae could not be removed from cocoons without problems with ecdysis. Pupae are easily distinguished from the preceding, larva-like pre-pupal stage by their resemblance to the adult shape, which can be seen from the opening of the cocoon (Fig. 1). Newly molted pupae were immediately removed from their cocoons and placed individually in a small (25 mm) weigh boat (VWR, Radnor, PA, USA) in the incubator (32 or 29 °C depending on the experiment, 60% RH, total darkness).

Pupal staging and photography

Pupae collected by the above described method were examined under the microscope (25 °C, 50% RH) every 4–6 h until adult ecdysis occurred. Morphological observations of different body parts, including the compound eyes (referred to hereafter as CE), face, legs, mesosoma, and metasomal segments, were documented across development to establish stages and their duration. Developmental staging was based on observations of 48 workers and seven males for B. impatiens, collected from five Koppert colonies, three workers and 20 males for B. vosnesenskii, collected from two colonies, and nine workers and 11 males for B. melanopygus, collected from five colonies. Defined stages were those that were the least variable and thus most informative for staging within and across species. The duration of each pupal stage was documented for B. impatiens workers.

Whole pupae as well as specific characteristics used as staging criteria were imaged. High resolution images of whole pupae were taken at each pupal stage using image stacking with a Canon 7D MarkII digital DSLR camera and Zeren Stacker software. Morphological criteria used for staging pupae were imaged with a Nikon Coolpix 4,500 digital camera attached to an Olympus SZ61 microscope using an eye piece adaptor (Optem 257014).

Examining setal pigmentation during pupal and early adult stages

To visualize temporal patterns of setal pigmentation during pupal stages, overlying pupal cuticle was removed from metasomal terga in Phosphate-buffered saline (PBS) to reveal the developing setae underneath. Images were taken immediately after dissection. Temporal patterns of setal pigmentation post-eclosion from the cocoon, otherwise referred to as the callow stage, were visualized for three species—B. impatiens, B. vosnesenskii, and B. melanopygus (red form) by imaging whole bees for different time periods post-eclosion (0, 6, 12, 24, and 48 h). To obtain callows, cocoons containing late stage pupae were kept in the incubator and checked every 4 h for adult eclosion. Newly emerged adults were isolated immediately, reared in a plastic box provided with sugar and pollen, and fresh-frozen (color does not change after freezing) at the appropriate stage.

Measuring temporal patterns of body melanization

The most variable feature across morphological stages of pupal development in bumble bees is pigmentation. To quantify the temporal pattern of pigmentation for each body part, newly molted pupae of B. impatiens were reared at 32 °C and 60% RH and imaged every 12 h until adult ecdysis. Images were transformed to gray scale in Image J and gray values were measured for different body parts, including the CE, head, mesosoma, legs, antennae, and metasoma, for each time point. In gray scale images, the gray value ranges from 0 (white) to 255 (black). To represent the increase of body pigmentation we calculate relative darkness as gray value of area tested/255 X 100%. One image of a whole pupa was used for each time point. We also estimated overall percent melanic body pigmentation by combining scores across all body parts except for the eye, which involves non-melanin pigmentation.

Measuring pupal duration

Previous studies tended to consider pupation as the period between cocoon formation and eclosion from the cocoon (Cnaani, Robinson & Hefetz, 2000; Sutcliffe & Plowright, 1990), which thus includes non-pupal stages of spinning larval and prepupal stages, and newly molted adult bees (Fig. 1). We examine the effects of sex, caste, temperature, and species on total pupal duration through observation of pupa reared under controlled environmental conditions outside the cocoon. We are thus able to demarcate pupal duration more accurately as the time between larva-pupa ecdysis (molting of larval cuticle) and pupa-adult ecdysis (molting of pupal cuticle). To measure pupal duration, newly molted pupae collected as above (accurate within 4 h) were left undisturbed in the incubator until the last pupal stage. At this stage, we checked the bees every 2–4 h to document the end of pupation, marked by the pupal cuticle being pushed to the posterior tip of the bee (Jay, 1965). Analysis of the effects of body size, caste, sex, and temperature on developmental duration were performed on B. impatiens, with individuals drawn from four Koppert (n = 83 individuals) and one wild queen-started colony (n = 24 individuals). Inter-tegular span (ITS), a common measure of bumble bee body size (Cane, 1987; Hagen & Dupont, 2013), was measured using a microscope eyepiece ruler (Olympus, Shinjuku, Tokyo, Japan). To avoid potential fluctuations of body size during pupal development, pupal sizes were determined by measuring ITS on newly eclosed adults once the adult ecdysis was complete. We designated three size groups for workers and males: small (ITS < 3.5 mm), medium (3.5 mm ≤ ITS < 4.0 mm), and large (ITS ≥ 4.0 mm). This grouping was chosen to represent the most meaningful by-eye groupings of typical bee sizes (intermediate) as opposed to those that appear atypically large or small. It matches well with size-based task groups in bumble bee workers recognized by Goulson et al. (2002) using B. terrestris, where small workers with ITS < 3.5 perform almost exclusively in-nest tasks, large workers with ITS > 4.9 are mostly foragers, and medium sized workers perform both tasks. Because natural bumble bee colonies maintain a nest temperature between 29 and 32 °C (Heinrich, 1972; Pomeroy & Plowright, 1980; Vogt, 1986a), we examined pupal duration for brood reared at 29 and 32 °C to represent the lowest and highest extremes for brood development. Measurements of pupal duration for B. impatiens at these temperatures included small worker (n = 18 for 32 °C, n = 7 for 29 °C), medium worker (n = 20 for 32 °C, n = 6 for 29 °C), large worker (n = 15 for 32 °C, n = 6 for 29 °C), small male (n = 7 for 32 °C, n = 3 for 29 °C), medium male (n = 13 for 32 °C), and queen (n = 8 for 32 °C, n = 8 for 29 °C) treatment groups. A Mann–Whitney test was performed to test temperature effects on pupal duration. To test if there are differences across phenotypes, a one-way ANOVA followed by Tukey’s HSD All-Pairwise comparisons were performed for the two temperature treatments separately. To compare between species, we measured pupal duration of small males (ITS < 3.5 mm) of B. vosnesenskii (one colony) and B. impatiens at both 32 and 29 °C (for B. vosnesenskii, n = 8 for 32 °C, n = 3 for 29 °C) and analyzed the data using a Mann–Whitney test.

Structures unique to pupae

Although mostly simply outlining the adult shape, the bumble bee pupal cuticle exhibits a few characteristics not shared with the adult stage. One obvious feature is the metasomal spicules, a row of small spines near the apical margin of metasomal tergites (Fig. 2A). In B. impatiens, the spicules are present at metasomal segments two through five (female) or six (for male). These spicules are present throughout pupal development and are shed along with pupal cuticle during pupal-adult ecdysis without replacement in the adult. These structures are absent in pupae of honey bees (Fig. 2B). Another feature is the presence of inner apical protuberances on the coxae and trochanters of the legs (Figs. 2C and 2D). These long protuberances house long setae arising in these segments (Fig. 2E) and appear to act as a brace for the developing legs (Fig. 2C). Honey bee pupae have the same structures at the same location. The coxal and trochanteral protuberances are most distinguishable at the early-mid pupal stage, but collapse and become indistinguishable before pupal-adult ecdysis (Figs. 2F–2H). Similar-shaped structures were also observed on tibial segments, but are retained into adulthood as tibial spines (Michener, 1954).

Pupal stages

We describe 20 pupal stages as well as two early post-ecdysis stages that are applicable to all three studied species and across sexes and castes. Key features for each stage are annotated using illustrations or photos (Figs. 3–5). The defined features primarily involve shifts in eye pigmentation in early stages (Fig. 3), followed by shifts in pigmentation of the developing adult cuticle of the main body parts (Fig. 5). Two distinct cuticle pigmentation processes were observed: a “peppering” process, in which the white cuticle acquires a grayish tinge, often starting as fine gray dots before darkening to black, and a “tanning” process, in which cuticle first turns a more even orange-brown color before turning black. Cuticle pigmentation is later followed by pigmentation of the adult setae, which continues until early adult (callow) stages (Fig. 6) Characteristics differing from B. impatiens in other examined species (B. vosnesenskii, B. melanopygus), or by caste or sex, are noted. Simplified morphological criteria for each stage are also provided in Table 1. A photographic guide of whole pupae of each stage featuring B. impatiens males is presented in Fig. 7.

General pigmentation patterns

A summary plot for timing of melanization on various body parts is provided in Fig. 8 and the relative progression of CE, cuticle (quantified as the sum of values from Fig. 8) and hair color is summarized in Fig. 9. These plots show very little overlap between eye and body pigmentation. Although pigmentation tends to intensify around the same time across non-eye body parts, the order is somewhat staggered (Fig. 8). Pigmentation is complete first in the mesosoma and head, followed by the metasoma, and then the appendages, including antennae and legs, which are not fully pigmented until closer to eclosion. Setae follow similar patterns to the main cuticle, although with a slight delay, with prospective black setae starting pigmentation with a light orange tinge after melanization of metasomal cuticle begins and intensifying gradually as pupae age, toward an orange-gray color that is maintained until around the time of adult ecdysis (Fig. 9). At eclosion, prospective black setae are orange-gray, red setae pink-gray, and yellow setae white (Fig. 6B). Color progression of black setae in B. impatiens and B. vosnesenskii and red setae of B. melanopygus begins immediately after eclosion, and is complete about 24–48 h later (Fig. 9). In contrast, pigmentation of yellow setae of all three species examined does not begin until 24 h after eclosion and completes at about 48–72 h (Figs. 6B and 9).

Pupal duration

Figure 10A compares pupal duration of different phenotype groups reared at 32 and 29 °C. Mean pupal duration and standard error are provided in Table S1. At 32 °C, pupal duration of workers ranges from 160 h (6.6 days, small workers 161.9 ± 1.5 h) to 180 h (7.5 days, large workers 179 ± 1.6 h), males take an average of 189 h (7.9 days, small males 188.6 ± 2.4 h, medium male 189.4 ± 1.7 h), and queens about 243.2 ± 2.2 h (∼10 days, Fig. 10A; Table S1). Pupal duration is significantly longer at the lower temperature for all groups (29 °C) (p < 0.01, Mann–Whitney test) (Fig. 10A; Table S1), with variation among groups in the magnitude of response to temperature. In particular, males and queens respond less than workers. Small, medium and large worker pupae reared at 29 °C take 20, 34, and 28 h longer to develop than similarly sized counterparts reared at 32 °C, respectively. However, small males reared at 29 °C take 15 h longer, and queen takes 22 h longer to develop than their 32 °C-reared counterparts (Fig. 10A and Table S1).

Overall the ANOVA reveals a strong effect of phenotype (size x caste) on pupal duration (32 °C, F _{5, 80} = 206.4, p < 0.0001; 29 °C, F _{4, 29} = 370.12, p < 0.0001, One-Way ANOVA). Body size affects pupal duration in the worker caste significantly, with larger workers taking 18 and 26 h longer to development than small workers at 32 and 29 °C, respectively (Tukey’s HSD post hoc test, with α = 0.05) (Fig. 10A). An effect was not found in males of different size (Tukey’s HSD post hoc test, with α = 0.05). There were also significant differences between castes at both rearing temperatures (Fig. 10A). In general, queens display substantially longer pupal duration than workers (Fig. 10A). For instance, at 32 °C, queen pupae take 81, 71, and 62 h longer to develop than small, medium, and large sized worker pupae, respectively (Tukey’s HSD post hoc test, with α = 0.05). Male pupae also take longer to develop than same-sized worker counterparts (Fig. 10A) with small sized males taking 27 and 22 h longer to develop than small sized worker pupae at 32 and 29 °C, respectively (Tukey’s HSD post hoc test, with α = 0.05). The effects of body size within caste also varies between temperatures. Significant differences were found between all three size groups within the worker caste at 32 °C (Tukey’s HSD post hoc test, with α = 0.05). However, medium-sized workers were no different from large workers at 29 °C (Tukey’s HSD post hoc test, with α = 0.05).

Body size does not alter developmental timing of all stages equally. It appears that the duration of developmental stages for small and medium B. impatiens workers were similar up until the orange-eyed pupal stage P5 (first 50 h) (Fig. 9). After this there was an increase in the timing of each stage with increased body size.

Due to limited availability of bees from B. vosnesenskii colonies at the time of experiments, only small males at 32 and 29 °C were used for comparison (Fig. 10B). B. vosnesenskii male pupae develop 10 h faster than those of similar size in B. impatiens at 32 °C (p = 0.0003), but no significant difference was found between the two species at 29 °C (p = 1.0).

Morphological transitions in pupal development

The current study provides the first comprehensive description of morphological changes during pupal development in bumble bees. As opposed to previous staging based mostly on eye color, the morphological criteria we established from multiple body parts enables staging of pupal development into smaller, more evenly distributed increments (Fig. 9). This morphological key, annotated with high resolution images, is intended for use in any study on developing bumble bees requiring precise pupal staging.

The characterization of bumble bee pupal stages presented here differs substantially from those described in honey bees and other bee species, with the exception of eye pigmentation (Elias-Neto, Soares & Bitondi, 2009). It includes more stages than defined in previous studies on bees, both in eye color and body pigmentation stages (Fig. 9 and Table S2). While the relative timing of cuticle melanization to eye pigmentation is largely consistent with observations in honey bees (Rembold, Kremer & Ulrich, 1980), the progression of melanization differs. This is to be expected given the pronounced difference between the species in their external morphology and pigmentation patterns at adult stages, and their large evolutionary distance (>90 my; Grimaldi & Engel, 2005). We observed very little difference, however, within and between bumble bee species in pupal traits, consistent with low variability observed within other genera (Fukutomi et al., 2017; Michelette & Soares, 1993). This key should thus be more broadly applicable across bumble bee species. That said, given that we focused only on a subset of clades within the bumble bee subgenus Pyrobombus, studies of non-Pyrobombus taxa may want to use this as a guideline for developing a more taxon-specific staging.

Our study highlights several features of pupal development in bumble bees. We first note the features unique to the pupal cuticle that differ from the pharate adult, notably apical tergal spicules and leg protuberances. Large across-species variation has been found in the pupal spicules in bees in terms of their number, size, positions and shapes (Baker, Kuhn & Bambara, 1985; Cardale, 1968; Michener, 1954; Rozen, 1990, 2000; Torchio & Trostle, 1986). We noted them to be pronounced in bumble bees but absent in honey bees. Although the function of these is unclear, given their morphology they may serve a purpose as mechanosensory setae or for anchoring while in the pupal cocoon, and thus may differ due to difference in their pupal environment. Pupal leg protuberances, also variably called spines (Michener, 1954) or tubercles (Rozen, 1990), may be relevant to setal development (Michener, 1954) given that they are filled with long setae, but appear to embrace the leg, thus may be holding the legs in place during development. We also observed the shift in pupal segments around the larval-pupal transition. Most non-sawfly Hymenoptera exhibit the synapomorphy of a constriction between their abdominal tergites 1 and 2 that leads to the flexible wasp waist separating the mesosoma from the metasoma. We provide a video (Video S1) observing the regression in these segments that recapitulates the progression of events involved in the formation of the wasp waist.

We have made several observations regarding the shifts in pigmentation across pupal development. We note that the eye follows a dot-to-hexagonal progression likely reflecting development of ommatidia that aligns with shifts in intensification of eye color (Fig. 3C). Eye color also appears to be decoupled from cuticle color, with the cuticle not pigmenting until eye color is complete. Eye color comes from ommochromes used as filtering pigments in the developing ommatidia, a different class of pigments from the brown-black cuticular pigments, which are melanin. Melanization was observed to start in body sutures, including legs joints and sutures on the mesosoma, and then tended to fill in first in the mesosoma, then the metasoma, followed by a progression of darkening from proximal to distal on the appendages. In the metasoma we observed striping patterns in these bees as cuticle melanizes that move posterior to anterior on metasomal tergites (Fig. 5). It is interesting to consider why melanin progresses as it does, with possibilities being related to sources of melanin precursors like tyrosine derived from the center of the body or the availability of oxygen required for melanization (Lerner & Fitzpatrick, 1950). Two melanization processes were observed to take place on the primary cuticle, some of which involve gray to black and others which start with a more ochre color, in line with alternative routes to black color in these bees noted by Milliron (1971). This could involve differences in melanin pathways, including differences between pheomelanin and eumelanin or other quinone derivatives (Hines et al., 2017). Finally, the yellow setal pigmentation is also fairly decoupled from melanin pigmentation, occurring a day or so after adult eclosion when melanization is mostly complete. This decoupling likely occurs because yellow setae are colored by a different type of pigment, likely a pterin (Hines, 2008).

Pupal duration

We observed shorter pupal duration in all size/caste groups when they were reared at higher temperatures. Both temperatures used are within the typical range of temperatures observed in the brood of bumble bee colonies (Heinrich, 1972; Pomeroy & Plowright, 1980; Seeley, 1981; Vogt, 1986a). Higher temperatures within a colony than outside are achieved by the activity of the adults and active thermoregulation of the brood through incubation (Heinrich, 1972; Seeley, 1981). This suggests that colony development will be impacted by incubation efforts. The negative correlation between rearing temperature and pupal duration is consistent with results from other insects (Elmes & Wardlaw, 1983; Foley, 1981; Kipyatkov, Lopatina & Imamgaliev, 2005; Penick et al., 2017). Interestingly, the worker caste is affected more than males and queens by temperature, with greatest effect on medium-sized workers, suggesting that developmentally plastic phenotypes like size and caste may have differential response thresholds to environmental conditions.

Our study reveals a positive correlation between body size and pupal duration within workers, with larger workers taking longer to develop than smaller ones. This could explain variation in worker pupal duration (up to 5 days) in previous studies (Katayama, 1973; Sakagami, Akahira & Zucchi, 1967). Given that pupal and adult size is determined at the onset of pupation from larval size, pupal duration is likely a consequence of body size rather than vice versa. In our observations, the timing of eye pigmentation phases did not differ by bee size but later stages involved in melanic pigmentation across the body did (Fig. 9). This may be because the eye is smaller and involves a single-layer (thus reduced complexity) (Cagan & Ready, 1989), while melanization would need to take place over a larger surface and may be impacted by signals coming from other parts of the body, which may take longer in larger individuals. The differences we find in both duration and potential lengths of specific stages by body size, means standardization of pupal stages (e.g., for genetic research) is best done on similarly sized individuals.

A positive correlation between body size and duration can fail when sexes and castes are considered. In Drosophila, male pupae are smaller than females but take longer to eclose (Nilssen, 2006; Nunney, 2007), and in the honey bee A. mellifera, queen pupae, which are larger than worker pupae, have a shorter pupal duration (Rembold, Kremer & Ulrich, 1980). Bumble bee queens do not follow the same patterns as honey bees, as queens have much longer pupal duration than workers and males. This matches expectations from body size, as queens are much larger than workers. The rationale for why a honey bee queen would take less time is interesting to consider. It could be due to their caste being determined more by diet quality (Kamakura, 2011). Alternatively, the contrast between bumble bees and honey bees may be explained by their life history: honey bee queens are under strong selection to develop quickly as the first queen to emerge typically becomes the new queen and kills the remaining queen-destined brood (Degrandi-Hoffman et al., 1998; Michener, 1974), a practice not exhibited by bumble bees which disperse and independently start their own nest. Bumble bee males, however, exhibited longer pupal duration than workers regardless of their body size, and the duration of their development was less sensitive to relative body size. Longer pupal duration in males may be because gonotrophic development is more complicated in males than the non-reproductive worker caste, thus requiring more time to complete metamorphosis (Breev, Zagretdinov & Minar, 1980). Different sexes and castes must therefore have adaptations driving duration that are independent of the effects of size.

We also found a difference between B. impatiens and B. vosnesenskii in pupal duration. Differences within and between closely-related species in developmental timing are not uncommon in insects (Elmes & Wardlaw, 1983; Fukutomi et al., 2017; Ma, Huang & Wang, 2015; Nilssen, 2006; Sgolastra et al., 2012), and, in bumble bees, differences in habitat have been suggested to select for different developmental rates (Sakagami, Akahira & Zucchi, 1967). These two species differ in their climatic niche, as B. vosnesenskii is a Pacific coastal bumble bee and has a longer season with milder and wetter weather than B. impatiens, which is exposed to eastern North American cool temperate, moderately wet seasonality. This could also explain observed differences in plasticity in these species, as the response to temperature in developmental duration was greater in B. vosnesenskii than in B. impatiens.